# Previous Colloquia

### Previous Colloquia Announcements

Monday, May 3rd, 2021,3:00 pm, Zoom Link: https://miami.zoom.us/j/96500582611

Mr. Prajwal Devkota

will present his PhD Thesis

Information flow and controllability of Biological Networks

Abstract

Functional processes and mechanisms of cells are not the result of a single molecule or component but rather tend to be related to complex and intricate interactions of these components. Hence, it is beneficial to study the complete system to understand the biological system in its entirety. Focusing on the flow of information through a molecular interaction network using a network propagation approach, we hypothesized that the difference in the flow of information between control and injured conditions prioritizes relevant driver nodes that cause this state change. We compare the difference in information flow in protein interaction networks between mice with axonal injury and sham surgery to identify causal nodes. We found kinases to be enriched among source genes sending significantly differential information to transcription factors, which we considered to sink because of their ability to regulate many genes. Such kinases were experimentally validated for their influence in axon growth based on either screening data or external literature confirmation.

Investigating cellular activity that drives the cell to specific outcomes, we determined proteins that topologically control different human pathways represented as independent molecular interaction networks. Transcending different topological levels, we show that proteins that control many pathways also control the network of interaction when all pathways are combined. Such control proteins were also found to be enriched with regulatory and signaling genes, disease genes, and drug targets suggesting biological significance.

We then introduce a new approach to identify controller nodes based on the edges of the network. We identify transcription factors that control different tissue-specific regulatory networks through enhancer and promoter edges. We show that target genes regulated by controller transcription factors through promoter edges are tissue-specific, overexpressed in cancerous tissues, and are likely to be targeted by cancer drugs in the corresponding tissue. We compare our method to a node-based approach and show that our method is better suited in identifying tissue specificity, leading to a better understanding of roles played by regulatory elements in regulation and disease at a tissue level.

Friday, April 2nd , 2021 3:00 pm, Zoom Link

Mr. Joseph G. Masterjohn
will present his PhD Thesis

Constructing the Arrangement of Algebraic Surfaces in a
Bounding Box

Abstract

The construction of arrangements is a fundamental operation in Computational Geometry. I present an
algorithm for calculating the arrangement of generic algebraic surfaces inside an axis aligned bounding
box. "Generic" means that inside the bounding box the surfaces are non-singular, intersect transversely,
and have no four-way intersections. We compute a data structure computed we call an Encasement, and
we show how to extract the arrangement components (vertices, edges, facets, and cells) and their topology
from this data structure. Common queries we provide are: computing a vertex to a specified accuracy,
sampling an edge or facet to specified accuracy, determining the cell containing a point, and for
generating a point on a facet. The algorithms are implemented robustly.  Input polynomials have double
precision floating point coordinates.  Calculations are done using double precision interval arithmetic with
ambiguous predicates evaluated using extended precision interval arithmetic.

Friday, April 2nd, 2021 2:00 pm, Zoom Link

Ms. Katarzyna Pasternak
will present her MS Thesis Defense

Reflective Embodied Conversational Agent for Human
Support Robot in Rapport Building - Pilot Study

Abstract

Building rapport in human-robot interaction is a key component for
facilitating communication between the user and intelligent autonomous
robotic platforms. To convey their intentions, both sides can use a variety of
verbal and nonverbal behaviors. In human-human interaction, one such
nonverbal behavior is partial subtle mirroring of the conversational partner's
nonverbal behaviors during conversation, which is also referred to as parallel
empathy. Furthermore, in regard to verbal behaviors, human interlocutors often
convey understanding via reflective listening by rephrasing their conversational
partner's words. While the effects of mirroring nonverbal behaviors and
reflective listening have both been studied on an individual basis in interactions
between robots and humans, as well as in interactions between Embodied
Conversational Agents (ECAs) and humans, there exist only a few studies that
examine interactions between humans and ECAs integrated with robots that
utilize such rapport-building techniques. The contribution of this thesis is a
pilot study on a prototype integration of an expressive ECA capable of
reflective listening, tracking its interlocutor's face, and mirroring their facial
expressions and head movements in real time with a Toyota Human Support
Robot such that the robot and the ECA are fully aware of and capable of
synchronizing with each other's and the users' verbal and nonverbal behavior.
This research, based on a pilot study, examines whether integrating the

aforementioned ECA with the support robot will improve the user's experience
while interacting with the support robot.

Friday, March 19th, 2021 2:00 pm, Zoom

Mr. Tong Liu
will present his PhD. Defense
Computational explorations of 3D genome structures - Resolution enhancement of bulk Hi-C
Abstract

High-resolution Hi-C data are indispensable for the studies of 3D genome organization at kilobase level. However, generating high-resolution Hi-C data (e.g., 5 kb) by conducting Hi-C experiments needs millions of mammalian cells, which may eventually generate billions of paired-end reads along with a high sequencing cost. Therefore, it will be important and helpful if we can enhance the resolutions of Hi-C data by computational methods. We developed a new computational method named HiCNN that used a 54-layer deep convolutional neural network to enhance the resolutions of Hi-C data. The network contains both global and local residual learning with multiple speedup techniques included for fast convergence. We used mean squared errors and Pearson's correlation coefficients between real high-resolution and computationally predicted high-resolution Hi-C data to evaluate the method. The evaluation results show that HiCNN consistently outperforms the state-of-the-art method HiCPlus when training and testing data are extracted from the same cell type and from two different cell types in the same or different species. We further present a deep-learning package named HiCNN2, an improved version of HiCNN. The HiCNN2 package includes three methods each with a different deep learning architecture: HiCNN2-1 is based on one single convolutional neural network (ConvNet); HiCNN2-2 consists of an ensemble of two different ConvNets; and HiCNN2-3 is an ensemble of three different ConvNets. Our evaluation results indicate that HiCNN2-enhanced high-resolution Hi-C data achieve smaller mean squared error and higher Pearson's correlation coefficients with experimental high-resolution Hi-C data compared with existing methods HiCPlus and HiCNN. Moreover, all of the three HiCNN2 methods can recover more significant interactions detected by Fit-Hi-C compared with HiCPlus and HiCNN.

Tuesday, 24th November 2020, 4:00pm, Zoom

Dr. João Sedoc
Department of Computer Science, John Hopkins University

will present

Evaluating Conversational Agents

There has been a renewed focus on dialog systems, including non-task driven conversational agents. Dialog is a challenging problem since it spans multiple conversational turns. To further complicate the problem, there are many possible valid utterances that may be semantically different. This makes automatic evaluation difficult, which is why the current best practice for analyzing and comparing dialog systems is the use of human judgments. This talk focuses on evaluation, presenting a theoretical framework for the systematic evaluation of open-domain conversational agents, including the usage of Item Response Theory (Lord, 1968) for efficient chatbot evaluation and evaluation set creation. We introduce ChatEval (https://chateval.org) a unified framework for human evaluation of chatbots that augments existing tools and provides a web-based hub for researchers to share and compare their dialog systems.

Friday, 20th November 2020, 3:00pm, Zoom

Mr. Tong Liu
Department of Computer Science, University of Miami

will present

Computational Explorations of 3D Genome Structures-bin-specific Embeddings of Single-cell Hi-C Data to Reconstruct 3D Genome Structures and Detect Topologically Associating Domains

In the last ten years, a lot of computational methods have been developed with population/bulk Hi-C data as input for modelling three-dimensional (3D) structures and identifying topologically associating domains (TADs) of chromatins. However, bulk Hi-C data are from pooling contacts over millions or billions of cells and cannot detect 3D structures of single cells. Moreover, increasing single-cell Hi-C data reveal cell-to-cell variance. Therefore, most of the existing methods are not applicable to single-cell Hi-C contact matrices. In this proposal, we present “scHiCEmbed”, an integrated framework for 3D structure reconstruction and TAD identification based on single-cell Hi-C data. scHiCEmbed consists of two main computational steps. The first step is the imputation on raw single-cell Hi-C data with an existing method scHiCluster. The second step is to embed each bin in a latent space using graph neural networks. For 3D structure reconstruction, we use two graph neural layers to encode bins in the 3D space: the first layer is to learn hidden representations in higher dimension; and the second layer’s task is to reduce the higher dimension to 3D. The learning processes are to reconstruct the input graph (raw or imputed single-cell Hi-C matrix) and to make any two bins with contacts as close as possible to each other and any two bins without contacts as far away as possible from each other simultaneously. For TAD identification, we simply use one graph neural layer as an encoder and an inner product to reconstruct input graph. The learned stochastic latent variable is passed to a constrained hierarchical clustering algorithm to hierarchically combine two sequentially successive bins or newly created clusters. S_Dbw is used to determine the optimal number of clusters, and we think of each cluster as a TAD. Average precision (AP) and area under ROC curve (AUC) are used for evaluating the quality of reconstructed input graph. Our testing results show that most of our models trained for the two tasks (3D reconstruction and TAD identification) can achieve AP and AUC  0.85. Using the reconstructed 3D structures, we plan to further explore the relationship between the compactness of our reconstructed 3D structures and cell division stages. For TAD identification, we will focus on the methylation densities around TAD boundaries and the overlapping of TAD locations detected from single-cell Hi-C data and bulk Hi-C data.

This is a Department of Computer Science PhD Proposal.

Monday, 24th August 2020, 3:00pm, Zoom

Mr. Pedro Peña
Department of Computer Science, University of Miami

will present

Trajectory Planning for Humanoid Robots and Inverse Trajectory Planning for Service Robots:
A Bridge Between Robots and Humans

In the near future, there will be robots that will live in our homes. The types of robots in our homes will not only be vacuum robots that are small and non-intrusive, there will also be general purpose robots that will be able to help us cook, do our laundry, and move around our homes in ways that resemble humans. Never in history have we ever had such agents, and they might impose danger if we do not carefully consider their behavior and motion planners. For a robot to move in our space, it needs to not only know how to avoid obstacles but also infer the position of a human in the environment. Furthermore, there needs to be an interface that allows robots and humans to communicate to understand what the robot is planning as well as for the robot to know what the human desires. Therefore we propose a vision JAMP, Joint Autonomy in Motion Planning, that consists of four components: Trajectory Planning, Inverse Trajectory Planning, Human-Robot Interface, and Robot Expressions, to produce a behavior that will consider the human’s desires as well as motions that take into account future human states to convey intention and safety to other humans. In my research, I worked on different components of JAMP individually to tackle different problems that are unsolved. I made contributions to trajectory planning for humanoid robots that is able to generate coupled motions while maintaining the robot stable. The trajectory planner was applied to soccer humanoid robots for kicking a ball while the robot is walking. The adaptive walk-kick trajectory controller for humanoid robots allows humanoids to kick in any direction while it is walking and was verified on a NAO robot; the stability and reliability of each kick has been evaluated 30 times for each kick motion trajectory while performing demanding motions. I also introduce a trajectory planner that considers humans in the environment in which a robot follows a human based on a nonlinear optimization to infer the 3D pose of a human wearing an infrared transmitter. Inverse trajectory planning allows a robot to infer future plausible states of a human, so I propose a novel probabilistic framework for robotic systems in which multiple models can be fused into a circular probabilitymap to forecast human poses, allowing models from different applications to integrate into one representation which can be used by the robot. We tested the framework on Toyota’s HSR robot and Waymo Open Dataset. Moreover, a human- robot interface that allows a human and robot to communicate is also considered in this research in which an augmented reality based human-robot interface that can be used to place and navigate a robot and map and label an environment will be introduced, as well as a human-robot interface that integrates multimodal communication features of a 3-dimensional graphical social virtual agent with a high degree of freedom service robot, HSR. We demonstrate HSR greeting gestures using culturally diverse inspired motions, combined with our virtual social agent interface, and we provide the results of a pilot study designed to assess the effects of our multimodal virtual agent/robot system on users’ experience. For robot expressions, I discuss philosophical remarks of how robots in the future will move and interact in the presence of humans and other robots and how they affect other agents in the environment. We think about how we can evaluate the effectiveness of expressions in agents and introduce a metric, Pena coefficient, that measures the potential of agent expression based on DoR, degrees of reality. Moreover, we revisit social agents and inspect how social actions can be seen from the perspective of degrees of freedom and present ASCEE and SDoF that can be used to measure social degrees of freedom in agents, a concept never visited as of yet, but is important to think about when evaluating agents in the environment of humans. We introduce human-robot laws that describe different types of interactions that will be deemed necessary in the future for human-robot interaction and present different levels of interactions between humans and robots and robots and robots. As we think about the interactions between these agents, we introduce a paradox, we call Robot Expression Paradox, which is introduced when robot-robot and human-robot expressions affect each other. To potentially resolve this paradox, we introduce REN, a robot expression network, that can be used to propagate changes among these different interactions.

This is a Department of Computer Science PhD Defense.

Wednesday, 4th March 2020, 5:05pm, MM217

Dr. Radka Stoyanova
Department of Radiation Oncology, University of Miami

will present

Deciphering Prostate Heterogeneity: Habitat Imaging

Screening for prostate cancer reduces prostate cancer mortality; however, it leads to oversampling, overdiagnosis and overtreatment of indolent cancer. Both randomized trials and long-term observational data support active surveillance (AS) as a safe alternative to immediate treatment in appropriately selected patients. However, as emerging data show an increased risk of metastasis for patients with delayed treatment, there are increasing concerns as to the proper selection of patients for AS. Prostate cancer'???'s multifocal and highly heterogeneous nature is the Achilles heel of effective patient risk stratification, and standard ultrasound-guided biopsies have proven to be inadequate as a measure of sampling for clinically significant prostate cancer. In imaging, this heterogeneity is mapped using the different sequences in Multiparametric MRI (mpMRI) of the prostate, or so called "habitats" - areas with different risk for aggressive cancer. We have developed a quantitative mpMRI pixel-by-pixel habitat risk score (HRS) platform, which is superior at identifying high-grade disease of Gleason score (GS) 3+4 and higher based on a prostatectomy cohort in ROC-AUC analysis. HRS is displayed as a 3D heat-map and implemented in the biopsy workflow to assure that the most aggressive habitat of the tumor is targeted.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217.

Wednesday, 26th February 2020, 5:05pm, MM217

Christopher Mader
Center for Computational Science, University of Miami

will present

Land Access for Neighborhood Development (LAND)Using open data and open source technology to help address affordable housing in Miami Dade County

Land Access for Neighborhood Development (LAND) is web-based mapping tool built to help visualize the distribution of local institutional, government-owned vacant and underused properties. Developed through a collaboration between the UM Office of Civic and Community Engagement and the UM CCS (IDSC) with support from Citi Community Development (Citibank), LAND enables policymakers and community-based organizations to identify potential affordable housing development opportunities in transit-served ares. A multi-tier application, LAND is built using a combination of Javascript (AngularJS), HTML, CSS (bootstrap), node.js., Apache Solr and PostGIS. The presentation will address both the need/usecase for LAND as well as its architecture.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217.

Wednesday, 19th February 2020, 5:05pm, MM217

Joseph Johnson
Miami Herbert Business School, University of Miami

will present

DEEPSENSE: A Deep Neural Network System for the
Analysis and Prediction of Ad Effectiveness

Current testing of ad effectiveness rely on methods such as content analysis, manual content annotation, focus groups, and opinion surveys. These methods are costly, time consuming and lack immediacy. We propose a computational approach that uses deep neural representations of visual, audio, and natural language content of ads to predict viewers' responses. Our approach employs the content as input and does not require the manual specification of content features as needed by conventional machine learning. We develop our approach on YouTube video ads and to build a predictive model of viewer response we use the YouTube engagement data (like/dislike, comment sentiments), as well as ratings by human coders. We find that of the three media data - video, audio and text - the video data is the most predictive of viewer response. This signifies that visuals contain the main predictive information for ad response. We also find that including computationally extracted multimedia timelines akin to the narrative timelines of stories improves the prediction of viewer response and that most multimedia timelines do not induce brand purchase intentions. The main implication of our work is that computerized systems can complement and perhaps even substitute current methods for providing viewer feedback on ad effectiveness.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217.

Wednesday, 12th February 2020, 5:05pm, MM217

Alberto Cairo
School of Communication, University of Miami

Scotney Evans
School of Education and Human Development, University of Miami

Barbara Millet
School of Communication, University of Miami

will present

Improving Hurricane Risk Communication For Vulnerable Populations:
Why We Misinterpret Hurricane Maps and What We Can Do About It

The National Weather Service, the National Hurricane Center, and other organizations regularly publish forecasts of where tropical storms and hurricanes may go and cause damage. However, a growing body of evidence suggests that most people can't read those visual products - maps, graphs, etc. This talk describes an effort to understand where misunderstandings of hurricane products may come from, and what we can do to address them.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217.

Wednesday, 5th February 2020, 5:05pm, MM217

Yelena Yesha, Ph.D.

Visiting Distinguished Professor, Department of Computer Science
University of Miami

Distinguished University Professor, Department of Computer Science and Electrical Engineering
University of Maryland, Baltimore County

Director
NSF Center for Accelerated Real Time Analytics

will present

Data Science for Medical Imaging

Artificial intelligence (AI) has great potential to augment the clinician as a virtual radiology assistant (vRA) through enriching information and providing clinical decision support. Deep learning is a type of AI that has shown promise in performance for Computer Aided Diagnosis (CAD) tasks. A current barrier to implementing deep learning for clinical CAD tasks in radiology is that it requires a training set to be representative and as large as possible in order to generalize appropriately and achieve high accuracy predictions. We present an Active Semi-supervised Expectation Maximization (ASEM) learning model for training a Convolutional Neural Network (CNN) for lung cancer screening using Computed Tomography (CT) imaging examinations. Our learning model is novel since it combines semi-supervised learning via the Expectation-Maximization (EM) algorithm with active learning via Bayesian experimental design for use with 3D CNNs for lung cancer screening. ASEM simultaneously infers image labels as a latent variable, while predicting which images, if additionally labeled, are likely to improve classification accuracy. The performance of this model has been evaluated using three publicly available chest CT datasets: Kaggle2017, NLST, and LIDC-IDRI. Our experiments showed that ASEM-CAD can identify suspicious lung nodules and detect lung cancer cases with an accuracy of 92% (Kaggle17), 93% (NLST), and 73% (LIDC) and Area Under Curve (AUC) of 0.94 (Kaggle), 0.88 (NLST), and 0.81 (LIDC). These performance numbers are comparable to fully supervised training but use only slightly more than 50% of the training data labels.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217.

Monday, 3rd February 2020, 11:15am, UB230

Dr. Ulas Bagci

University of Central Florida

will present

Robust and Explainable Machine Learning for Radiological Systems

Vision researchers have been analyzing behaviors of radiologists during screening to understand how and why they miss tumors or misdiagnose. In this regard, eye-trackers have been instrumental in understanding visual search processes of radiologists. However, most relevant studies in this aspect are not compatible with realistic radiology reading rooms. In this talk, I will share our unique experience for developing a paradigm shifting computer aided diagnosis (CAD) system, called collaborative CAD (C-CAD), that unifies CAD and eye-tracking systems in realistic radiology room settings. In other words, we are creating artificial intelligence (AI) tools that get benefits from human cognition and improve over complementary powers of AI and human intelligence. We first developed an eye-tracking interface providing radiologists with a real radiology reading room experience. Second, we proposed a novel computer algorithm that unifies eye-tracking data and a CAD system. The proposed C-CAD collaborates with radiologists via eye-tracking technology and helps them to improve their diagnostic decisions. The proposed C-CAD system has been tested in a lung and prostate cancer screening experiment with multiple radiologists. More recently, we also experimented brain tumor segmentation with the proposed technology leading to promising results. In the last part of my talk, I will describe how to develop AI algorithms which are trusted by clinicians, namely “explainable AI algorithms". By embedding explainability into black box nature of deep learning algorithms, it will be possible to deploy AI tools into clinical workflow, and leading into more intelligent and less artificial algorithms available in radiology rooms.

Wednesday, 29th January 2020, 5:05pm, MM217

Dr. Orlando Acevedo

Department of Chemistry
University of Miami

will present

Simulating Ionic Liquids: Development and Applications

Due to the negative environmental impact of damaging organic solvents and the high-cost of chemical waste disposal, the search for alternative, renewable solvents is a top priority in the chemical industry. Part of a worldwide emphasis on improving sustainability, the push has led to increasing interest in a set of environmentally-friendly solvents: ionic liquids (ILs). These solvents, primarily composed of ions, e.g., [RMIM][PF6], provide numerous advantages from both chemical and environmental perspectives. Despite the excitement surrounding these solvents, a fundamental understanding of how they operate is still lacking. This seminar will highlight our efforts developing and applying computer models to both elucidate how these solvents function and to design the next generation of cost-effective ???green??? solvents. Specific topics to be covered include force field parameterization, the development of machine learning neural network software for simulating chemical reactions, and the creation of machine learning genetic algorithms for automated force field development.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217.

Wednesday, 22nd January 2020, 5:05pm, MM217

Dr. Gang Ren

Center for Computational Science
University of Miami

will present

Scaling Up Heterogeneous Waveform Processing for Long-Duration
Monitoring Signal Acquisition, Analysis, and Interaction:
Bridging Big Data Analytics with Measurement Instrument Usage Pattern

Modern oscilloscopes, data loggers, and related electronic measurement instruments generate a large amount of waveform data for many long-duration waveform capturing and analysis tasks. The contrast of time scales of long-duration waveform capturing (e.g., signal length of hours in high sampling rate) and analysis (e.g., signal fragments of several microseconds) produces unique big data challenges for signal acquisition, storage, organization, analysis, and user interactions. We have designed and implemented several long-duration waveform processing algorithms for waveform data organization, analysis, visualization, and user interaction at big-data scale. To cope with the real-time processing and user interaction demands within the hardware constrains of instrument platforms, the proposed algorithms utilize multiple layers of test signal synthesis, data pre-sorting, database query, and time series similarity quantification at various speed-precision tradeoffs. We also integrated typical instrument usage patterns into the design process to provide intuitive user interfaces for efficient data organization, flexible data record navigation, and in-depth experimentation.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217.

Thursday, 9th January 2020, 11:15am, UB230

Dr. Jia Zhang

Carnegie Mellon University - Silicon Valley

will present

Facilitating the Reuse of Data Analytics Capabilities

Modern big data analytics are typically constructed through a multi-step procedure, often referred to as a "workflow", with each step in a workflow representing a specific algorithm or function. Over the years, many scientists have been frequently recreating their own data analytics algorithms from scratch. This results in a major duplication of effort, and it wastes money, time and other important resources. In this talk, I will discuss the necessity of building a data science infrastructure capable of recommending reusable algorithm resources (artifacts) when scientists design big data analytics workflows. I will illustrate an application of machine learning and information retrievalmethods to tackle the automatic annotation problem. Aiming to highlight the uniqueness of an artifact, automatic annotation is an important problem in data science, since descriptions from contributors might be subjective and not kept up to date. Our work illustrates the use of a generative model that learns from artifact usage provenance to refine artifact representations, and uses these learnings to support a category-aware artifact recommendation method.

Tuesday, 10th December 2019, 10:00am, UB230

Mr. Md Nasir Uddin Laskar

Department of Computer Science
University of Miami

will present

Understanding Texture Sensitivity in the Brain Visual System
with Deep Neural Networks

Deep convolutional neural networks (CNNs) trained on visual object recognition have shown intriguing ability to predict some response properties of brain visual systems. However, the factors (e.g., if the model is trained or not, receptive field size etc.) and the computations (e.g., convolution, rectification, pooling, normalization etc.) that give rise to such ability are yet to be understood. Considering texture sensitivity as a paradigmatic example, and recent brain experimental data in the second cortical visual area (V2), I will show some qualitative similarities between the CNN model neurons and the brain data across a number of metrics. I will also present a systematic quantification technique that I developed, to select neurons from the CNN layers that best describe the brain experimental data. This research opens the door to answer a long standing research question about correspondence between primate and machine vision.

This is a Department of Computer Science PhD Defense.

Thursday, 5th December 2019, 12:30pm, UB305

Dr. Kuang-Ching Wang

Clemson University

will present

Networking Research in an Era of Convergence

Networking research has played a central role in a broad range of major research pursuits in the last two decades. Networking has been a key driver for advancements from HPC to cloud, from big data to artificial intelligence (AI), from Internet of Things (IOT) to cyber physical systems (CPS), from precision medicine to precision agriculture, and from smart cities to connected communities. Throughout this exciting period, I have been fortunate to participate in a series of national-scale NSF projects aimed at discovering, enabling, and transforming research and stimulating economy across broad science and engineering disciplines. In many ways, our efforts contributed to the now prevalent “convergence” strategy in numerous major NSF and NIH programs. My research group focuses on networking and computer systems research set in application contexts from genomics to health care, transportation, smart power grid, autonomous vehicle, and a broad range of AI projects. The unique research trajectory also helped me appreciate the power of “team science”. In this talk, I will give an overview of my research journey, the impressive collaborations I have had with people in U.S. and globally, and share my thoughts on how networking research will continue to be a core strategy for transformative advances in science, engineering, and our society.

Monday, 2nd December 2019, 11:15am, UB230

Dr. Zhizhong Han

Department of Computer Science
University of Maryland, College Park

will present

Deep Learning based 3D Shape Understanding

Understanding 3D shapes is vital in 3D computer vision, especially as unmanned systems become ever more connected in our life. The overarching goal is to propose intelligent algorithms to help machines to understand the 3D would like what humans can do in various applications, such as classification, recognition, segmentation, shape completion, and shape captioning, et al, which however is still challenging. In this talk, I will present my recent works on 3D shape understanding using novel deep learning models. Specifically, I will introduce how we can leverage supervised or unsupervised methodology to enable deep learning models to learn from 3D shapes in different representations, such as multiple views and point clouds. The introduced deep learning models are discriminative or generative networks, which are designed to finish different tasks in different applications. Based on my experience, I will also present some new directions or methods that would improve the ability of 3D shape understanding.

Wednesday, 20th November 2019, 5:05pm, UB230

Prof. Liang Liang

School of Architecture
University of Miami

will present

Towards Robust Convolutional Neural Networks for
Malaria Parasite Identification in Thin Blood Smear Images

The detection of malaria is conventionally performed by laboratorians through visually examining blood smear images under a microscope. Convolutional neural networks (CNNs) have great potential for automated image-based malaria diagnosis without human interaction. However, CNNs could be vulnerable to adversarial noises, which have been demonstrated on some popular image datasets, such as ImageNet: state-of-art CNNs make wrong predictions for input images with adversarial noises. In this study, we developed a CNN for malaria parasite identification, with the goal of improving CNN robustness to adversarial noises.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar.

Monday, 18th November 2019, 11:15am, UB230

Dr. Vanessa Aguiar-Pulido

Weill Cornell Medicine
Cornell University

will present

AI for Data-Driven Knowledge Discovery in Human Disease Research

The advent of high-throughput sequencing technologies has ushered in a new era of genetic inquiry, decreasing the economic costs and turnaround time notably. However, with this growth of ‘omic’ data, computational challenges arise. Not only the volume of data is increasing but also the dimensionality, expanding the search space exponentially. As a result, many problems involve such a large set of possible solutions that finding the optimal one in a reasonable amount of time is not feasible. Therefore, developing new approaches for ‘intelligent’ data analysis is necessary. This talk will describe how artificial intelligence has been applied to identify potential genetic risk factors in complex genetic disorders (i.e., those that require genetic predisposition and environmental factors for the patient to develop the disease). Starting from an evolutionary computation-based algorithm for candidate gene analysis, we will move toward a totally unbiased, comprehensive approach that uses machine learning (more specifically, embedded feature selection) in ethnically diverse cohorts. Approaches such as the latter are fundamental to further human disease research in the age of precision medicine.

Wednesday, 13th November 2019, 5:05pm, UB230

Chris Chung

School of Architecture
University of Miami

will present

RAD: Augmented Reality, Spatial Computing and Architecture

RAD-UM provides resources and expertise for project-based research on the spatial ramifications of embedded technology and ubiquitous computing. The research is premised on the notion that every building or landscape component can be equipped with computational power.Projects at RAD-UM develop models for such digitally enhanced environments to better handle persistent and emerging challenges in the areas of healthcare, building technology and sustainability.The projects are set up for multi-disciplinary collaboration and for potential development in partnership with industry. Computing is migrating from dedicated static appliances to mobile devices, objects of everyday life, and physical environments thanks to increasingly proliferating microchips and ever-expanding information networks.The spatial nature of ubiquitous computing directly implicates and empowers architecture, landscape, and urban design. RAD-UM capitalizes on this potential, bringing research to bear on the built environment from a variety of fields that exploit the spatial consequences of distributed computing responsive and interactive systems, augmented reality, embedded/situated technology, ambient intelligence, mobile computing and locative media.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar.

Wednesday, 6th November 2019, 5:05pm, UB230

Dr. Leo Siqueira

Department of Atmospheric Sciences
University of Miami

will present

High-performance Computing for Climate Modeling

This presentation will provide a brief overview of challenges in simulating weather and climate and the role of high-performance computing. Climate modeling requires representing a multitude of processes on a variety of timescales and space scales. It is also characterized by sensitive dependence on initial conditions and natural stochastic variability, so very long and ensemble integrations (many realizations showing the range of future climate outcomes) are needed to extract climate signals and estimate uncertainties. The increase in resolution, ensemble size, and inclusion of finer-scale processes in climate simulations leads to better accuracy but present particular high-performance computing challenge, taking current storage and computing resources to their limits.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar.

Wednesday, 30th October 2019, 5:05pm, UB230

Max Cacchione

Director of Innovation, UMIT
University of Miami

will present

Mixed Reality at UM

In this seminar, I will give an overview of current projects and an exploration of Mixed Reality modalities, and show interactive learning demos.

This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar.

Wednesday, 23rd October 2019, 5:05pm, UB230

Prof. Dilip Sarkar

Department of Computer Science
University of Miami

will present

Blockchain Technology and its Applications

As of October 2019, total market capitalization of cryptocurrencies is about $223 billion. To create Bitcoin, blockchain technology was combined with several other technologies. Now blockchain technology is used for many digital ledgers for diverse applications. In this talk, an overview of blockchain technologies is presented. First, background and history of technologies that are foundation of tamper evident and resistant digital ledgers are discussed. Two categories of blockchains and their current and potential future applications will be presented. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar. Wednesday, 9th October 2019, 5:05pm, UB230 Prof. Lindsay D. Grace School of Communication University of Miami will present Making Games at Record Speed Attendees will learn the basic tenets of making functional, publishable games in less than 7 days. The talk draws from the speaker's experience as Vice President of the non-profit that runs the largest game jam in the world, as well as personal experience as a top 100 mobile game developer. The talk combines practical tips with heuristics refined over making scores of digital games, professional consulting, and years of leading students toward successful game designs. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar. Wednesday, 2nd October 2019, 5:05pm, UB230 Mr. Amr Elsawy Department of Electrical and Computer Engineering University of Miami will present Imaging Processing and Computer Vision Methods for Diagnosis of Eye Disease Optical Coherence Tomography (OCT) is a technology that images the front and back parts of the eye(i.e. cornea and retina) using backscattered light. OCT images show different layers of the parts of the eye. Different eye diseases change the shape of these layers which is an important sign for the presence of eye diseases. Imaging processing and computer vision methods are needed for the extraction of clinically relevant parameters from the OCT images which is necessary for the diagnosis of these diseases. These methods include image registration, image segmentation, 3D reconstruction, correction and thickness measurement. Our work focuses on processing corneal OCT images. We are building a corneal layer-thickness topography (CLTP) system to generate 3D thickness maps of different corneal layers. These maps are used for diagnosing different eye diseases such as keratoconus, Fuchs endothelial dystrophy and corneal graft rejection. In this talk, we will explore the different parts of our system. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar. Friday, 27th September 2019, 11:15am, UB230 Dr. Yelena Yesha Computer Science and Electrical Engineering University of Maryland Baltimore County will present Chios: Next Generation Permissioned Blockchain Blockchain, widely known for being the technology underlying Bitcoin, has started to creep into areas that extend beyond cryptocurrency. Blockchain is a secure and reliable distributed service that maintains digital records in a tamper-proof way. There are two types of blockchains: permission-less blockchains (e.g., Bitcoin, Ethereum) and permissioned blockchains (e.g., Hyperledger Fabric). Existing blockchain systems are still facing several major challenges such as fine-grained access control, better support of messaging pattern, and modularity in design. In this talk, I will introduce Chios, a permissioned blockchain system that is unique in multiple applications such as IoT and data sharing. Chios features novel techniques to address significant challenges in blockchains. First, Chios is the first-ever system that achieves decentralized confidentiality and decentralized cryptographic based fine-grained access control. Second, Chios enables publish/subscribe mechanisms while resolving open problems in distributed systems, achieving strong security guarantees. Last but not least, Chios is truly modular and extensible. It currently supports more than 30 configurations, allowing various trade-offs among efficiency, function, and security. Also, Chios provides visualized interfaces so one can easily deploy and use Chios services and applications. Dr. Yelena Yesha is a tenured Distinguished University Professor at the Department of Computer Science and Electrical Engineering at the University of Maryland, Baltimore County. She is also the Director of the National Science Foundation Center for Accelerated Real Time Analytics (CARTA). She received her B.Sc. degrees in Computer Science and in Applied Mathematics from York University, Toronto, Canada, in 1984, and her M.Sc. degree and Ph.D. degree in Computer Science from The Ohio State University in 1986 and 1989, respectively. She has published 11 books as author or editor, and more than 170 papers in prestigious refereed journals and refereed conference proceedings, and has been awarded external funding in a total amount exceeding 40 million dollars. Dr. Yesha is currently working with leading industrial companies and government agencies on new innovative technology in the areas of cybersecurity and big data analytics with applications to electronic commerce, climate change and health. Dr. Yesha is a fellow of the IBM Centre for Advanced Studies. This is another in the Department of Computer Science Distinguished Scientist Seminar series. Wednesday, 25th September 2019, 5:05pm, UB230 Clay Ewing, Ph.D. Department of Cinema and Interactive Media University of Miami will present The Importance of Design In today's society we constantly interact with systems. Whether designing a game, software application, database or anything else, it's important to design with purposeful intent and understand the impact that we are creating. This talk will cover mistakes made, lessons learned, and some hot tips on design. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar. Friday, 20th September 2019, 11:15am, UB230 Nick Tsinoremas, Ph.D. Center for Computational Science University of Miami will present Data Driven Compute Systems: A Case Study in Bioinformatics and Healthcare Data Science The Biomedical Informatics architecture at UM currently supports nearly 7 PB of biomedical research data and numerous projects including EHR data from three Health Care partner organizations. This session will provide an overview of the University of Miami Center for Computational Science's (UM CCS) data driven advanced computing systems design as well as the design of the new AI-ready system, Triton which support the biomedical architecture. The overall Biomedical Informatics and Bioinformatics architecture and the Miami CTSI program will be discussed in detail along with new innovative data driven approaches and applications for Health Care and Biomedical Research. Also included in this session will be an overview of University Research Integration Data Environment (U-RIDE) which is a core component of this architecture comprised of three main layers: URIDE Foundation, URIDE Analytics, and URIDE Search. Wednesday, 18th September 2019, 5:05pm, UB230 Dr. Zheng Wang Department of Computer Science University of Miami will present SCL: A Lattice-based Approach to Infer Three-dimensional Chromosome Structures from Single-cell Hi-C Data In contrast to population-based Hi-C data, single-cell Hi-C data are zero-inflated and do not indicate the frequency of proximate DNA segments. There are a limited number of computational tools that can model the three-dimensional structures of chromosomes based on single-cell Hi-C data. We developed SCL (Single-Cell Lattice), a computational method to reconstruct three-dimensional (3D) structures of chromosomes based on single-cell Hi-C data. We designed a loss function and a 2D Gaussian function specifically for the characteristics of single-cell Hi-C data. A chromosome is represented as beads-on-a-string and stored in a 3D cubic lattice. Metropolis-Hastings simulation and simulated annealing are used to simulate the structure and minimize the loss function. We evaluated the SCL-inferred 3D structures (at both 500 kb and 50 kb resolutions) using multiple criteria and compared them with the ones generated by another modeling software program. The results indicate that the 3D structures generated by SCL closely fit single-cell Hi-C data. We also found similar patterns of trans-chromosomal contact beads, Lamin-B1 enriched topological domains, and H3K4me3 enriched domains by mapping data from previous studies onto the SCL-inferred 3D structures. The C++ source code of SCL is freely available at http://dna.cs.miami.edu/SCL/. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar. Wednesday, 11th September 2019, 5:05pm, UB230 Donald B. Olson, Ph.D. Department of Ocean Sciences University of Miami will present Multi-variate Approaches to Fisheries Habitat Fisheries, the act of taking fish, depend upon the physical and biotic factors that set up the habitat of targeted species and their availability to the fishery itself in terms of technology and proximity to market. At the same time, most of the hard data on these fish comes from the fishery catch itself making reconstruction of habitats and management of fisheries difficult. Setting up a system for defining fishery habitat from catch data and other available information is explored. In particular new fishery-independent data from sources such as tagging, genetics, and satellite mapping of ocean environments are discussed. Exploratory models of fish habitat are introduced and their use in fisheries described. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar. Wednesday, 4th September 2019, 5:00pm, UB230 Gang Ren Center for Computational Science University of Miami will present Multimodal Grammatical Inference from Musical Expressions: Learning Syntactic, Semantic, and Structural Representations Towards A Systematic Musical Language Musical expressions carry rich structural and semantic information; our musical percept embraces multimodal experiences: visual, auditory, lyrics, scores, and myriad facets of artistic and intellectual concepts. Music’s expression is cognitively straightforward but analytically (reasoning rationally or “formally”) elusive: we all can genuinely feel the music, but a systematic decoding of musical expressions, a hypothetical parallel formal modeling process like that we have achieved on natural language processing, is still beyond reach. This presentation will report our initial attempts and progresses toward forming a systematic musical language using grammatical inference from multimodal representations of music. Specifically, we utilize a machine translation paradigm that models different but synchronized feature modalities as semantically coupled “languages”. From there, we systematically explore the syntactic, semantic, and structural properties of musical data. We will bring you along an amazing artistic adventure from Palestrina to Beethoven to Debussy and to Motown. We will compose a more rebellious Camila Cabello with Steve Reich’s tutelage, render an inappropriate cantata for a royal wedding toward impressionism and experimental, and spatially mixing Dr. Dre (instrumental only) with Stockhausen. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar. Wednesday, 28th August 2019, 5:00pm, UB230 Matthew Vega-Sanz Co-founder/CEO of Lula Rides will present From 1 to 10: Building the Technical Team Beyond Myself Matthew is the technical founder, and will discuss the journey of growing the Development Team beyond just himself. He will talk about problems we encountered while building and scaling the app, as well as the problems we had while hiring. Then, he will discuss how we overcame that, and still are working to overcome this. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar. Wednesday, 21st August 2019, 5:00pm, UB230 Dr. Michael Mannino Center for Computational Science University of Miami will present Computational Modeling of Brain Networks I will describe and show a neuroinformatics package called The Virtual Brain, which simulates the human brain, using computational models to explore the dynamics of large scale brain networks, and how information flows between different brain regions.I will also discuss some more abstract issues about the use of simulation in computer science, and neuroscience in particular. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in the reception area of the 3rd floor of Ungar. Wednesday, 24th April 2019, 5:00pm, MM217 Dr. Amin Sarafraz Center for Computational Science University of Miami will present Multi View Stereo: New Developments Multi-view stereo algorithms tackle a classic computer vision problem: reconstructing 3D geometry of objects using a set of images. The 3D output has applications in mapping and navigation, online shopping, 3D printing, augmented and virtual reality, video games, cultural heritage documentation, to name a few. This talk briefly describes the multi view stereo problem and provides an overview of some established and successful MVS pipelines as well as some new developments in this field. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 17th April 2019, 5:00pm, MM217 Dr. Farid Rajabli Hussman Institute for Human Genomics University of Miami will present Use of Data Mining Approaches to Estimate the Genetic Ancestry in Human Disease Research The genetic make-up of admixed populations is not homogenous. Admixture creates mosaic chromosomes of distinct ancestry that lead to a variation of allele frequencies. Due to this structure, admixed populations may have more than one ancestral background across the samples at a particular genomic region. This could contribute to the cohort-level test differences often seen in association studies. For example, the strongest risk gene identified for late-onset Alzheimer disease is ApoE. However, the risk for Alzheimer disease due to ApoE is not consistent across the populations. Individuals with the African ancestry experience less risk from ApoE than individuals of European or Asian ancestry. Thus genetic ancestral methods may play an essential role in understanding the factors contributing to the lower/higher risk effect across the populations by using admixed populations. In particular, identification of population-specific variation that influences disease could inform precision medicine initiatives, and lead to the development of ancestry-specific disease treatments. This would improve treatments, and help reduce health disparities. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 10th April 2019, 5:00pm, MM217 Dr. Mei-Ling Shyu Department of Electrical and Computer Engineering University of Miami will present Deep Latent Representation Learning using Variational Autoencoders and Variational Inferencing Deep Learning puts forward some of the most sophisticated machine learning solutions to the current state-of-art Artificial Intelligence (AI) applications including natural language processing (NLP), audio processing, computer vision, etc. In spite of all recent achievements of deep learning, it has yet to achieve semantic learning required to reason about the data. This lack of reasoning is partially imputed to the boorish memorization of patterns and characteristics from millions of training samples and ignoring the underlying, often hidden, relationships. This is the reason why mere rotations or color-inversions can easily confound very deep neural networks even though the modified images share the same semantics and structures as the original images. Over the last decade, the Database Data-mining and Multimedia (DDM) research lab has been extensively working on Knowledge Representation and Reasoning (KRR) and reasoning based deep learning problems that focus on bridging the semantic gap between low-level features and their high-level contextual meanings. We believe that the next generation of AI systems will need to have the ability to understand problems at a deeper level rather than just based on memorization of data. This talk is about one of the core methods, called Variational Autoencoders, that we use to uncover the latent relationships in the observed data. Variational autoencoders work on the principle of variational inferencing that has gained lots of attention as being the models of choice for generative models. Variational autoencoders help us uncover deep representations by mapping the observed data to latent spaces and use the latent relationships and generative modeling to infer hidden information about data. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 28th March 2019, 5:00pm, MM217 Dr. Ching-Hua Chuan Department of Cinema and Interactive Media University of Miami will present Machine Learning in Human-Centered Research Recent advances in machine learning and artificial intelligence have dramatically redefined our everyday experience. From medical diagnosis to criminal justice, machine learning algorithms start to play an important role in our life. Unlike most machine learning applications that focus solely on data, this presentation discusses human-centered research to build personalized, interdisciplinary, creative applications. In this presentation, I will present my projects in three domains: computational music research, natural language processing and American Sign Language recognition, and machine learning for wellbeing. I will also briefly discuss the courses that I teach, including Augmented Reality and Design with AI. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 21st March 2019, 5:00pm, MM217 Dr. Timur M. Urakov (M.D.) Jackson Memorial Hospital University of Miami will present Improving Augmented Reality Experience for Neurosurgery As Augmented Reality is gaining popularity in all fields of life, the specialized field of brain and spine surgery is sure enough to be one of the first to take advantage of this incredible technology. Imaging modalities that are currently used in the operating room are expensive, bulky, and produce harmful radiation. AR has the potential to alleviate this burden and early studies have shown its feasibility in doing so. However, there is still a tremendous amount of work that needs to be done. Current AR systems are not completely optimized for the operating room use. Collaboration and understanding between the end-user operators and the developers will ensure a productive growth of the technology. Closing the feedback loop and identifying caveats that need improvement needs to be a part of every discussion. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 14th March 2019, 5:00pm, MM217 Dr. Gang Ren Center for Computational Science University of Miami will present Automatic Head Movement Tracking, Analysis, and Interpretation for Autism Spectrum Disorder Diagnosis Clinicians and researchers studying children for autism spectrum disorder (ASD) have long noticed the atypical head postures and movement patterns of children with ASD. They have used the difference in the head movement patterns between the children with ASD and those without ASD as supplementary diagnostic cues based on manual interaction and analysis. Our work provides an automatic head movement tracking, analysis, and interpretation framework for ASD diagnosis and research. Specifically, we implemented a computational framework for extracting the temporal patterns of head movement and utilizing the imbalance of temporal pattern distribution between diagnostic categories (children with or without ASD) as potential diagnostic cues. The temporal patterns are extracted from multiple motion feature dimensions and time resolutions to form a high-dimensional "big-data" feature array. Then our proposed framework identifies the temporal patterns with imbalance distributions between contrasting diagnostic categories as potential clinical diagnostic cues. Our proposed automatic framework aims to improve clinical operational efficiency while allowing human experts to focus on essential diagnostic decisions or to tackle datasets of larger scales. The proposed analysis and interpretation framework is also useful for exploring the motor movement processes of ASD-related atypical head movement for their translational integration towards therapeutic interventions. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 6th March 2019, 5:00pm, MM217 Dr. Will Wei Sun Department of Management Science University of Miami will present Personalized Advertising and Ad Clustering via Sparse Tensor Methods Tensor as a multi-dimensional generalization of matrix has received increasing attention in industry due to its success in personalized recommendation systems. Traditional recommendation systems are mainly based on the user-item matrix, whose entry denotes each user's preference for a particular item. To incorporate additional information into the analysis, such as the temporal behavior of users, we encounter a user-item-time tensor. Existing tensor decomposition methods are mostly established in the non-sparse regime where the decomposition components include all features. In online advertising, the ad-click tensor is usually sparse due to the rarity of ad clicks. In this talk, I will discuss a new sparse tensor decomposition method that incorporates the sparsity of each latent component to the CP tensor decomposition. In theory, in spite of the non-convexity of the optimization problem, it is proven that an alternating updating algorithm attains an estimator whose rate of convergence significantly improves those shown in non-sparse decomposition methods. The potential business impact of our method is demonstrated via an application of click-through rate prediction for personalized advertising. In the second part of the talk, I will discuss an extension of the proposed sparse tensor decomposition to handle multiple sources of tensor data. In online advertising, the users' click behavior on different ads from multiple devices forms a user-ad-device tensor, and the ad characteristics data forms an ad-feature matrix. We propose a unified learning framework to extract latent features embedded in both tensor data and matrix data. We conduct cluster analysis of advertisements based on the extracted latent features and provide meaningful insights in linking different ad industries. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 27th February 2019, 5:00pm, MM217 Dr. Daniel Messinger Department of Psychology University of Miami will present Early Dyadic and Group Interaction: Leveraging Big Behavioral Data Developmental psychology is concerned with transformative change processes that occur over the human lifespan. Much early development occurs in the context of social interaction. New sensing technologies, combined with machine learning, may provide insight into the social and emotional development of typically developing children and those with communication disorders such as autism and deafness. Automated detection of smiling and vocal turn-taking is shedding light on the diagnosis of autism. Automated analysis of classroom movement and vocal interaction is suggesting how language development occurs among peers. The talk will consider the strengths, challenges, and future of objective measurement and modeling of child behavior. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 20th February 2019, 5:00pm, MM217 Dr. Weizhao Zhao Department of Biomedical Engineering University of Miami will present Fiducial-less Real Time Tracking in Radiation Treatment for Liver Tumors Medical Physics integrates computer science, mathematics, physics, engineering and medicine into an interdisciplinary profession in serving diagnostic imaging, radiation therapy, nuclear medicine imaging, and radiation control/protection. This presentation provides an application example of image-guided radiosurgery that demonstrates medical physics research in cancer treatment. Tracking tumor movement during the treatment is crucially important for radiation therapy. The gold standard of real-time tracking of abdominal tumors requires the use of fiducial markers, however, which usually induces complications. We hypothesized that the two-dimensional (2D) location of the lung-diagram border can be used to determine the three-dimensional (3D) location of the tumor with clinically acceptable accuracy in comparison to using fiducial markers. Using a simple two-layer ANN architecture, we built correlation models that link the lung-diaphragm border's location with the corresponding 3D location of the tumor volume. Both simulation study (proof of concept) and clinical study (validation of concept) show the feasibility of accurately predicting the tumor volume's position with the use of widely available kV imagers through machine learning without fiducial markers. This patented innovative technique has the potential to eliminate fiducial markers in the tracking of liver or other abdominal tumors. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 13th February 2019, 5:00pm, MM217 Dr. Jason S. Nomi Department of Psychology University of Miami will present MRI Approaches for Investigating Brain Function and Structure In this talk I will first provide an overview of cognitive neuroscience approaches by comparing and contrasting various invasive and non-invasive cognitive neuroscience techniques. I will then talk about using MRI to map large-scale structural and functional networks of the brain and how such approaches may be used to quantify developmental changes and identify differences in clinical conditions such as autism spectrum disorder. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 6th February 2019, 5:00pm, MM217 Dr. Gecheng Zha Department of Mechanical and Aerospace Engineering University of Miami will present Simulation in Aerospace Engineering Enhances Technology Revolutions In the past 5 decades, rapidly increasing of computing power and development of computational fluid dynamics enable virtual simulations of various aerospace systems. Such simulations make many proof of concepts must faster at a much lower cost. This seminar will talk about the impact that high performance computing brings to aerospace engineering and the future perspective. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 30th January 2019, 5:00pm, MM217 Dr. Weiyong Gu Department of Mechanical and Aerospace Engineering University of Miami will present Simulations of Degenerative Intervertebral Disc Disease and ClinicalTrials for Disc Repair Intervertebral disc (IVD) is the largest avascular structure in the human body and its main function is to support mechanical loading and to provide the flexibility for the spine system.Degenerative disc disease (DDD) is related to low back pain which affects more than 600 million people worldwide. One of the challenges in modeling DDD is that biological, chemical, electrical, and mechanical events in IVD are nonlinearly coupled. It is important to understand the biophysics and pathophysiology in IVD in order to develop a model successfully. A multiscale and multi-physics model for IVD has been developed recently. In this model, nonlinear interactions among biological (cell activity), chemical (osmolarity and pH), electrical (charges on matrix and solutes), and mechanical (loading and tissue swelling) events in the IVD are considered. Numerical results are obtained by solving a dozen of partial differential equations using a finite element method. Applications of this model to simulating degenerative progression of IVDs due to poor nutrition supply will be presented. In-silico clinical trials of cellular therapies for dis repair will also be discussed. The study provides not only new insights into the mechanisms of disc degeneration, but also new diagnostic means for disc degeneration. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Friday, 25th January 2019, 10:00am, UB330D Mr. Pedro Peña Department of Computer Science University of Miami will present An Omni-directional Kick Engine for the NAO Humanoid Robot Incorporating a dynamic kick engine that is both fast and effective is pivotal to be competitive in one of the world'???'s biggest AI and robotics initiative: RoboCup. Using the NAO robot as a testbed, we developed a dynamic kick engine that can generate a kick trajectory with an arbitrary direction without prior input or knowledge of the parameters of the kick. The trajectories are generated using cubic splines, sextic polynomials, or cubic Hermite splines, and are executed while the robot is dynamically balancing on one foot. When the robot swings the leg for the kick motion, unprecedented forces might be applied on the robot. To compensate for these forces, we developed a Zero Moment Point (ZMP) based preview controller that minimizes the ZMP error. Although a variety of kick engines have been implemented by others, there are only a few papers on how kick engine parameters have been optimized to give an effective kick. Parameters such as kick configuration, limit of the robot, or shape of the polynomial can be optimized. We propose an optimization framework based on the Webots simulator to optimize these parameters. Experiments of the physical robot show promising results. This is a Department of Computer Science MS Defence. Wednesday, 23rd January 2019, 5:00pm, MM217 Mr. Haluk Damgacioglu Department of Industrial Engineering University of Miami will present A Dynamic Data-driven Framework for Smart Operations Management of Smart Grids This presentation will discuss our three-layer dynamic data-driven application systems-based framework for the control and planning of smart energy systems. We will highlight the issues of network topologies in smart grids and then analyze the operations management of smart energy systems. Next, we will provide the details of our multi-fidelity simulation model for timely monitoring and system assessment. Last, we will briefly discuss the real-time decision-making module that adapts the operation plan from the optimization model using robust demand response, network reconfiguration and/or operation rescheduling models based on dynamically changing data from the smart grid. We will explain the results of the investigated framework based on our computational tests on IEEE-9, IEEE-30, IEEE-118 bus systems and the designed smart grid for the City of Coral Gables. The presentation will conclude with a discussion on the future for the smarter grids. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm outside MM217. Wednesday, 16th January 2019, 5:00pm, MM217 Mr. Jiří (George) Pavelka Department of Computer Science University of Miami will present Software Development Automation Using Conceptual Git Branching Automation means reliability. Reducing human factor in software development increases project's overall quality. In this presentation we will introduce ways to increase automation on 7 different levels iteratively by adding conceptual git "lanes" (branches). It is a concept, that can be applied widely regardless the size of the project, number of programmers, language or environment. The only question is how much automation and thus reliability does your project need. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30pm in MM217. Wednesday, 28th November 2018, 5:00pm, UB506 Chloe Arluck Department of Computer Science University of Miami will present Approximate Free Space Construction and Maximum Clearance Path Planning for a Four Degree of Freedom Robot Constructing a robot's free space can be useful for fast and accurate navigation, but construction is difficult in high dimensions. In this talk I will present an algorithm for closely approximating the free space of a robot with 4 degrees of freedom (for example, a drone helicopter). Additionally, I will show how the free space can be utilized for safe and reliable navigation. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 14th November 2018, 5:00pm, UB506 Dr. Jihye Bae Department of Biomedical Engineering Florida International University will present Development of Neural Decoders and Neural State Emulators towards Practical Brain Machine Interfaces There have been promising advances in brain machine interfaces. However, many challenges still remain before this technology can become practical. In this talk, I will discuss the main issues along with possible approaches to overcome them. In particular, I will introduce methods to translate neural signals to control external devices using reinforcement learning and a neural signal emulator to help accelerate brain machine interface research. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 7th November 2018, 5:00pm, UB506 Terry Dyess, Dr. Daniel LaCombe, Anthony Margetic 3GIMBALS will present PISCES Automated Maritime Port Activity Monitoring -- Cruising into Client Needs Discovery The commercial remote sensing sector has experienced a period of rapid growth over the past decade. High-frequency, medium-resolution (3-5 meter GSD), electro-optical satellite imagery presents a unique and unprecedented opportunity to perform daily monitoring over wide areas. The 3GIMBALS team will discuss their PISCES project. PISCES is an automated maritime port activity monitoring application designed to unlock the potential of this new data source for intelligence analysts in the defense sector. First, the team will outline the thought process and approach taken throughout the project including how they used agile software development, minimally viable products, and client-need discovery to move fast and delight the client. Next, they will walk the audience through the computer vision algorithm development, the types of time series analysis and anomaly detection algorithms applied on top of it, and the hardware-cloud-web hybrid infrastructure we use to deliver the service. Finally, they will discuss past and present challenges, as well as, share insights about potential future directions of the application into multi-source data fusion. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 31st October 2018, 5:00pm, UB506 Dr. Rafael Nepomechie Department of Physics University of Miami will present Introduction to Quantum Computing An international race is currently underway to develop quantum computers and quantum algorithms. In this elementary introduction, we'll address such questions as: What is it? Why is it important? And where are we now? This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 24th October 2018, 5:00pm, UB506 Joseph Weller OSE Strategic Outreach Program Manager will present Open Source Enterprise - Turning Open Source into Intelligence The Open Source Enterprise (OSE) is under CIA's Directorate of Digital Innovation (DDI) and is the Agency's newest Directorate focused on accelerating innovation across the Agency's mission activities with cutting-edge digital and cyber tradecraft and IT infrastructure. Please join us to learn about the history, mission, and career opportunities in Computer Sciences, of this unique organization. Representatives from CIA and OSE will be available to answer your questions. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 17th October 2018, 5:00pm, UB506 Dr. Daniel McGibney School of Business Administration University of Miami will present Predicting the Stock Market Stock data is abundant and all around us.It is the goal of many organizations, groups and individuals to know the direction and movements of the overall stock market as well as for individual stocks. In this talk, I'll discuss how to calculate technical indicators from historical stock data, and how to create features and targets out of historical stock data. I will then discuss several machine learning algorithms which can be used to predict the future price of stocks. And finally, the resulting performance of these models will be evaluated and compared. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 10th October 2018, 5:00pm, UB506 Dr. Gang Ren Center for Computational Science University of Miami will present Statistical Signal Pattern Categorization, Integration, and Interpretation in Deep Neural Network Based Multimedia Content Analysis Systems Large-scale deep neural networks usually employ sophisticated internal signal propagation structures and parameter settings and thus the learning process and results are difficult to interpret and to interact with manual processes. We propose a novel pattern interpretation and interaction framework that combines the predictive accuracy of deep learning systems and the pattern/process transparency of the conventional statistical pattern recognition algorithms in the application context of content-based prediction of manual multimedia annotation labels. Our implementation of deep neural network based multimedia content analysis system employs convolutional neural network layers for modeling the video frame pixels and the audio spectrogram samples. Then the feature maps from different video segments are combined using recurrent neural networks for predicting the annotation labels. The proposed pattern interpretation and interaction framework feeds in the statistical signal patterns into the feature fusing and temporal modeling layers of the deep neural networks. Experiments show that the deep neural networks tuned to specific suboptimal configurations receive the highest amount of boost of the predictive analysis performance from the feature integration process of appending external multimedia signal patterns, while these signal patterns alone do not have meaningful predictive performance on the same predictive analysis tasks. The performance enhancement pattern of integrating the external statistical patterns also helps to establish the information dependency structures between the video/audio signal patterns to the predictive targets of manual annotation labels and forms cognitively relevant interpretations of the multimedia signal patterns. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 3rd October 2018, 5:00pm, UB506 Dr. Liang Liang Department of Computer Science University of Miami will present Model Based Biomedical Image Analysis In this seminar I will present my research on model-based biomedical image analysis. In order to extract information of objects from biomedical images (e.g. microscopy images of cells), mathematical models were built to represent the objects and describe their properties. Strategies and algorithms were developed to fit the object models to image data, and then information about the objects was obtained. I will show three applications including subcellular particle image analysis, cell image analysis, and aortic valve image analysis. I will focus on the models, strategies and algorithms in these applications. The advantage of model-based approaches is that the models are interpretable and could be physics-based, and the disadvantage is that it may be time-consuming to derive and use such models. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 26th September 2018, 5:00pm, UB506 Dr. Miroslav Kubat Department of Electrical and Computer Engineering University of Miami will present Machine Learning. Where did we start and where are we going? After decades of languishing in university labs, machine learning has become respectable. Major companies have established entire divisions specializing on machine learning software, newspapers publish articles about spectacular applications, and the public is intrigued. A scientist who has spent a lifetime working in this field rejoices---and recapitulates. This is the time to pause, take a deep breath, and look back at the story of the discipline. What motivated its beginnings? Which major avenues has it followed? How has it conducted its business? Just as well, this is also a time to look ahead. Widely publicized success stories may have obscured some other benefits that machine learning can offer, beyond the now-so-popular recognition of complicated patterns. The talk will therefore briefly mention some unfairly neglected possibilities that may inspire future research. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 19th September 2018, 5:00pm, UB506 Mr. Tong Liu Department of Computer Science University of Miami will present Reconstructing High-resolution Chromosome Three-dimensional Structures by Hi-C Complex Networks Hi-C data have been widely used to reconstruct chromosomal three-dimensional (3D) structures. One of the key limitations of Hi-C is the unclear relationship between spatial distance and the number of Hi-C contacts. Many methods used a fixed parameter when converting the number of Hi C contacts to wish distances. However, a single parameter cannot properly explain the relationships between wish distances and genomic distances. We addressed one of the key issues of using Hi-C data, that is, the unclear relationship between spatial distance and the number of Hi-C contacts, which was crucial to understanding significant biological functions, such as the enhancer-promoter interactions. We developed a new method to infer this converting parameter and pairwise Euclidean distances based on the topology of the Hi-C complex network. The inferred distances were modeled by clustering coefficients and multiple other types of constraints. We demonstrated the methodology on biological data and reconstructed a high-resolution 3D chromosomal structures of mouse male cells. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 12th September 2018, 5:00pm, UB506 Dr. Luis Gonzalo Sánchez Giraldo Department of Computer Science University of Miami will present Deep Neural Networks: Incorporating Computations Inspired by the Brain Deep Convolutional Neural Networks have become the current state of the art in Computer Vision problems such as object and scene recognition. Most of their success has been attributed to the ability of these networks to learn a representation of the objects rather than using handcrafted features. Interestingly, these networks, which have been trained to perform high-level tasks, also display an intriguing ability to explain some aspects of visual neural processing in the brain. In our work, we exploit the representations learned by Deep Convolutional Networks to expand our understanding of what is currently known for primary visual cortex in the brain, to intermediate stages of the visual cortex. I will discuss how this work can have mutual impacts to inform models of the brain and to advance artificial intelligence. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 5th September 2018, 5:00pm, UB506 Mr. Hassen Dhrif Department of Computer Science University of Miami will present Enhanced Multi-objective PSO Algorithms for Bioinformatics Applications We deal with multi-objective optimization problems (MOOP) in our everyday lives. Some problems such as identifying the smallest subset of genes with the highest prediction accuracy of a genetic disease are simply not solvable in reasonable (polynomial) time and resort to metaheuristic approaches. Particle swarm optimization (PSO) is a simple, yet powerful metaheuristic evolutionary algorithm that has been successfully used in the last twenty years to solve complex MOOP. While powerful, such algorithms suffer from early and suboptimal convergence, leaving large areas of the search space unexplored. In this talk, I introduce COMB-PSO, a heuristic that solves such issues and further allows the minimization of the number of features and the enhancement of the convergence rate, precision, robustness and general performance. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 29th August 2018, 5:00pm, UB402 Dr. Dilip Sarkar Department of Computer Science University of Miami will present Blockchain and Its Applications As of August 22nd 2018, total market capitalization of crypto currencies is about$208 trillion. Out of that Bitcoin's share is about 111.2 trillion. This talk introduces blockchain, the technology behind cryptocurrencies. Satoshi Nakamoto proposed Blockchain in 2008 to serve as the public transaction ledger of the cryptocurrency bitcoin. Blockchains that are readable by the public is widely used for cryptocurrencies. A block in a blockchain contains cryptographic hash of the previous block, a timestamp, and data. We will discuss algorithms for mining' blocks. We will also discuss other potential applications of blockchain. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 25th April 2018, 5:00pm, UB230 Dr. Geoff Sutcliffe Department of Computer Science University of Miami will present Making Belnap's "Useful 4-Valued Logic" Useful In 1977 Nuel Belnap published two articles, "How a Computer Should Think" and "A Useful Four-Valued Logic", in which he defined a four-valued logic called FDE (First Degree Entailment). However, FDE does not allow entailments within statements, and no conditional connective is defined. As such, it is not really computationally "useful". This work proposes conditional connectives to add to FDE, and describes the implementation of a reasoning tool for FDE with a conditional connective, with experimental results. With the addition of a conditional connective FDE starts to become truly computationally useful. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 18th April 2018, 5:00pm, UB230 Dr. Vladimir N. Uversky Department of Molecular Medicine University of South Florida will present Dancing Protein Clouds: Intrinsically Disordered Proteins Intrinsically disordered proteins (IDPs) lack stable tertiary and/or secondary structure under physiological conditions in vitro. They are highly abundant in nature and have functional repertoire which is very broad and complements functions of ordered proteins. Often, intrinsically disordered proteins are involved in regulation, signaling and control pathways. Functions of IDPs may arise from the specific disordered form, from inter-conversion of disordered forms, or from transitions between disordered and ordered as well as between ordered and disordered conformations. The choice between these conformations is determined by the peculiarities of the protein environment, and many IDPs possess an exceptional ability to fold in a template-dependent manner. These proteins are often key players in protein-protein interaction networks being highly abundant among hubs. Regions of mRNA which undergo alternative splicing code for disordered proteins more often than they code for structured proteins. This association of alternative splicing and intrinsic disorder helps proteins to avoid folding difficulties and provides a novel mechanism for developing tissue-specific protein interaction networks. IDPs are tightly controlled in the norm by various genetic and non-genetic mechanisms. Alteration in regulation of this disordered regulators are often detrimental to a cell and many IDPs are associated with a variety of human diseases such as cancer, cardiovascular disease, amyloidoses, neurodegenerative diseases, diabetes and others. Therefore, there is an intriguing interconnection between intrinsic disorder, cell signaling and human diseases. Pathogenic IDPs, such as ?-synuclein, tau protein, p53, BRCA1 and many other disease-associated hub proteins represent attractive targets for drugs modulating protein-protein interactions. Several strategies have been elaborated for elucidating the mechanisms of blocking of the intrinsic disorder-based protein-protein interactions. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 11th April 2018, 5:00pm, UB230 Lindsay Thomas Department of English University of Miami will present The WhatEvery1Says Project: Machine Learning and the Humanities This talk will give an overview of the use of computational methods of text analysis in literary studies, focusing on a project I co-direct: the WhatEvery1Says project (WE1S). WE1S uses methods in computational text analysis to study public discourse about the humanities at large data scales, focusing on journalistic news media from about 1981 to the present. Our hypothesis is that digital methods can help us learn new things about how news media sources portray the humanities. For example, are there sub-themes beneath the familiar dominant clich?s and memes about the traditional value of the humanities? How do different parts of the world or different kinds of sources compare in the way they think about the humanities? How do mainstream media position people from different racial, ethnic, gender, first-generation-student and other groups relative to the humanities? How do media articles by or addressed specifically to such groups compare with mainstream media in how they depict this relationship? In the process of our work, we are also developing tools and guidelines to create an open, generalizable, and replicable digital humanities methodology - one based on standards that the data-intensive sciences have advanced under the rubrics of "open science" and "open lab. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 4th April 2018, 5:00pm, UB230 Michelle Afkhami Department of Biology University of Miami will present The Unseen Majority: Linking Microbiome Effects on Host Fitness to Genome-wide Expression, Coexpression Network structure, and Nonadditive Natural Selection Despite being invisible to the naked eye, microbes are everywhere. Not only do all plants and animals play host to diverse assemblages of microorganisms, these hidden players often provide critical benefits necessary for their host's survival. A key goal of molecular biology has been to understand the genomic basis of how these microbes benefit their hosts, but has focused on pairwise association between a host and a single microbe which ignores the reality that organisms interact with many microbes simultaneously. While ecological literature has embraced this complexity and demonstrated synergistic, non-additive effects of multiple microbes on host fitness, it cannot determine the mechanistic and genomic basis of how organisms respond to these interactions. In this talk, I will combine ecological experimental designs with genomic and computational tools to provide insight into the molecular basis of synergistic effects of microbiomes on host plant performance. In particular, I will show how two important plant-associated symbionts - mycorrhizal fungi and rhizobia - interactively influence: (1) genome-wide expression using gene-at-a-time approaches and enrichment analyses, (2) molecular interactions among genes using differential coexpression network analyses, and (3) non-additive selection on host fitness using genotypic selection analyses. By taking an integrative approach, this work collectivity provides a new perspective on the ecologically-important interaction between plants and their microbiomes. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 28th March 2018, 5:00pm, UB230 Sean Kenny CIO, Carnival Cruise Line will present Tech Talk in the Sea Suite with Carnival's CIO On any given day, Carnival Corporation has over 100 ships sailing worldwide, and one of our newest ships, Carnival Horizon is scheduled to make her maiden voyage this spring! We pride ourselves on delivering memorable vacations to our guests and our ships features day and nighttime entertainment like stage shows, musical performances, casinos and more. Our guests feel right at home, too - think comfy stateroom accommodations, attentive service, delicious food and drink - and they enjoy the experience against the backdrop of some of the world's most beautiful ports. Today we bring high tech to the high seas to transform our cruise ships into seaborne smart homes. Our teams are passionate about introducing new technology onboard the ships to truly enhance the cruising experience, and redefine what it means to have a personal 24/7 concierge onboard. Join us at this Pizza Seminar as Chief Information Officer of Carnival Cruise Line, Sean Kenny gives an executive's perspective on how our team truly enhances our guest experience, and how he launched his career in a digital world. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 21st March 2018, 5:00pm, UB230 Dr. Julia Dallman Department of Biology University of Miami will present Computer-assisted Phenotypic Analyses in Zebrafish Models of Autism Recent breakthroughs from high-throughput sequencing in humans, genome engineering in animals, genetically encoded calcium sensors to capture brain dynamics, to machine learning to quantify phenotypes/symptoms are providing unprecedented insights into the molecular genetics of brain function. The recent advances in genomic engineering have allowed our lab to collaborate with labs in human genetics to recapitulate human neuropsychiatric conditions in zebrafish models as a means to delineate underlying mechanisms. In my talk, I will discuss how computer-assisted analyses of phenotypes in our zebrafish models of autism are helping us to understand the biological mechanisms associated with negative symptoms that range from sleep disturbance and epilepsy to gastrointestinal distress. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 7th March 2018, 5:00pm, UB230 Dr. Lynn Perry Department of Psychology University of Miami will present Using Automated Measures to Examine Language Experiences in Early Intervention Classrooms Although parent speech has become a subject of increasing study and intervention, less is known about speech in childcare settings. What is known often comes from observations at the level of classroom rather than at the level of individual infant. For example, teacher language quality is assessed as a static trait of the teacher, missing the fact that classrooms are busy interactive places in which different infants have different types of conversations with each other and with their teachers. It is only with the use of automated measurement and big behavioral data that we can understand how different infants' experiences vary even within a shared classroom and what consequences that variability can have on their development. I will present data from two studies utilizing automated measures of language and movement in early intervention classrooms. Together, the results of these studies suggest that children's interactions with peers and teachers have cascading consequences for their future language development. Our application of automated measurement provides new insight into the dynamics of the classroom environment and consequences for language development in at-risk and delayed infants - insights that could not be obtained without the power of big data. Refreshments will be served. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 28th February 2018, 5:00pm, UB230 Dr. Diana K. Ter-Ghazaryan Department of Geography University of Miami will present Isn't that Spatial?: Maps and the Power of Location Geospatial technology is increasingly relevant to our lives in the 21st century. Join Dr. Ter-Ghazaryan, Lecturer and Director of Geospatial Technology Program at the Department of Geography, for an interactive presentation on the role that maps and Geographic Information Systems (GIS) play in our everyday lives. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Monday, 5th March 2018, 12:15am, Richter Library Conference Room 343 Dr. Liang Liang Georgia Institute of Technology will present Biomedical Image Analysis - Analyzing the Images of Bio-Particles, Cells, and Human Heart Valves I will present my research on biomedical image analysis which includes modeling, simulation and inference. My research focuses on the development of new algorithms and methods to resolve challenges in this field by utilizing techniques in computer vision, machine learning, and biomechanics. I will start from biological image analysis for studying cellular processes, including fluorescence imaging, object detection and tracking. The images from just one experiment may contain tens of thousands of objects (e.g. bio-particles) with high density, which makes it almost infeasible to manually analyze the image data. To resolve this challenge, I developed a novel multiple hypothesis based tracking method using state space representation and mixed integer programming, which automatically recovers the object trajectories and outperforms traditional methods. From the tracking results, we were able to infer the role of a chemical compound from the cellular behaviors. Then I will present my recent work on medical image analysis, especially image-based patient-specific modeling of human aorta and heart valves. Soft tissue biomechanical modeling and finite element (FE) simulation have shown promising results for medical diagnosis (e.g. evaluating thoracic aortic aneurysm) but have limited clinical translatability due to heavy manual workload and long simulation time. To resolve the bottlenecks, I developed machine learning based methods that can automatically build FE models from images and serve as a fast surrogate of FE simulation for stress analysis. I will also show machine learning based methods for soft tissue microscopy image analysis and lumbar spine image analysis. This is another in the Department of Computer Science Seminar Series. Wednesday, 21st February 2018, 5:00pm, UB230 Brian McNoldy Rosenstiel School of Marine and Atmospheric Science University of Miami will present Ensemble Techniques Applied to Ocean Profiling Floats in an Observing System Simulation Experiment Large ensembles of trajectories can be simple to generate, but very challenging to turn into a meaningful, digestible, and actionable end product. In the atmospheric and oceanic sciences, examples include ensembles of tropical cyclone forecast tracks over several days and ensembles of float forecast positions over several years. This paper specifically addresses the latter, but the techniques described apply to both. The Southern Ocean Carbon and Climate Observations and Modeling (SOCCOM) project is a major multi-year and multi-institutional effort to better understand the complex biogeochemical processes in the Southern Ocean and their influence on the global climate system. Observations in that hostile part of the world are sparse and expensive, so a high-fidelity numerical model simulation was created to mimic the real world as closely as possible. This simulation, or "nature run", is the foundation of an Observing System Simulation Experiment (OSSE) which is designed to help guide and inform decision-makers and optimize physical resources. Using the nature run, we introduce thousands of synthetic floats at specified locations along realistic cruise routes; those floats are then allowed to drift with the ocean currents for years. We employ several methods to analyze and visualize the resulting track ensembles, including trajectory density, target area probability, and data depth. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Monday, 19th February 2018, 12:15am, Cox 42 Mr. Hongfu Liu Northeastern Universityi will present From Consensus Clustering to K-means Clustering Consensus clustering aims to find a single partition which agrees as much as possible with existing basic partitions, which emerges as a promising solution to find cluster structures from heterogeneous data. It has been widely recognized that consensus clustering is effective to generate robust clustering results, detect bizarre clusters, handle noise, outliers and sample variations, and integrate solutions from multiple distributed sources of data or attributes. Different from the traditional clustering methods, which directly conducts the data matrix, the input of consensus clustering is the set of various diverse basic partitions. Therefore, consensus clustering is a fusion problem, not a traditional clustering problem. In this talk, I will introduce the category of consensus clustering, illustrate the K-means-based Consensus Clustering (KCC), which exactly transforms the consensus clustering problem into a (weighted) K-means clustering problem with theoretical supports, talk about some key impact factors of consensus clustering, extend KCC to Fuzzy C-means Consensus Clustering. Moreover, this talk includes how to employ consensus clustering for heterogeneous, multi-view, incomplete and big data clustering. Derived from consensus clustering, a partition level constraint is proposed as the new side information for semi-supervised clustering. Along this line, several interesting applications based on the partition level constraint, such as feature selection, domain adaptation, gene stratification are involved to demonstrate the extensibility of consensus clustering. Some codes are available for practical use. This is another in the Department of Computer Science Seminar Series. Friday, 16th February 2018, 12:00am, Cox 42 Mr. Zhengming (Allan) Ding Northeastern Universityi will present Knowledge Transfer for Face Recognition It is essential to mimic human cognitive process to adapt previous well-learn knowledge to facilitate the new challenging face recognition tasks. In this presentation, I focus on two topics, i.e., missing modality and one-shot learning. First of all, we may always confront the problem that we have no target face available in the training stage, which arises when the face data are multi-modal. To overcome this, we first borrow an auxiliary database with complete modalities, then propose a two-directional knowledge transfer to solve the missing modality issue. Secondly, we may always confront that we only have one training sample for some persons. It is challenging for existing machine learning approaches to mimic this way since limited data cannot well represent the data variance. To this end, we propose to build a large-scale face recognizer, which is capable to fight off the data imbalance difficulty. Specifically, we develop a novel generative model to synthesize meaningful data for one-shot classes by adapting the data variances from other normal classes. This is another in the Department of Computer Science Seminar Series. Wednesday, 7th February 2018, 5:00pm, UB230 Craig Pottinger Chief Technical Officer and Co-Founder of Graupel LLC will present Leveraging AR & ML to find your Personalized Fit Retailers and Ready to Wear brands have largely survived by inconveniently coercing customers to purchase generically sized apparel. However, there aren't enough sizes to categorize the full range of body shapes that people have. The unfortunate result is customers struggle to find clothes that uniquely fit them. The problem of improper fit has directly led to billions of dollars in apparel returns each year. At Graupel, we are putting the customer first. Graupel is an omnichannel apparel brand that leverages technology to ensure customers get their personalized fit. In this presentation with Craig Pottinger - Chief Technical Officer and Co-Founder of Graupel - he examines how they provide a unique fit to their customers. He dives into the technology stack that enables Graupel's business model and examines the current challenges around building an automated pipeline - a pipeline that starts with a 3D avatar of the customer and ends with a pattern made custom to the customer's body shape. Furthermore, he explores how various Machine Learning and Augmented Reality techniques would improve automation, quality control and conversion rates. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 31st January 2018, 5:00pm, UB230 Alex Rubinsteyn The Icahn School of Medicine at Mount Sinai will present Algorithmic Selection of Therapeutic Cancer Vaccines This talk will discuss computational methods used to design personalized therapeutic cancer vaccines. Therapeutic cancer vaccines, unlike preventative prophylactic vaccines, aim to mobilize an immune response against an established cancer. Since each patient's cancer contains a unique pattern of mutations, a different vaccine must be made for each patient. The starting point of this personalization process is DNA sequencing of both the patient and their tumor. The differences between the patient/tumor genomes reveal a pool of potentially targetable mutations, which are then further filtered by computational predictions of how mutations affect proteins and statistical models of which mutations can be recognized by the immune system. The aim of this talk is to demonstrate the very direct role which computational methods can play in a patient's cancer treatment. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 24th January 2018, 5:00pm, UB230 Dr. Shouraseni Sen Roy Department of Geography and Regional Studies University of Miami will present Geovisualizing Miami through Space and Time Geovisualization is used to refer to visualization techniques used for analyzing and presenting large geospatial datasets. This presentation showcases the results of various geospatial research questions based on Miami through space and time. It involves the use of multiple years of data and spatial datasets at different scales. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Thursday, 14th December 2017, 2:00pm, UB230 Mrs. Hande Kucuk-McGinty Department of Computer Science University of Miami will present Knowledge Acquisition and Representation Methodology (KNARM) and Its Applications Technological advancements in many fields have led to huge increases in data production, including data volume, diversity, and the speed at which new data is becoming available. In accordance with this, there is a lack of conformity in the ways data is interpreted. In-depth analyses - making use of various data types and data sources, and extracting knowledge - has become a more daunting task. This is especially the case in life-sciences where simplification and flattening of diverse data types often leads to incorrect predictions.Effective applications of big data approaches in the life sciences require better, knowledge-based, semantic models that are suitable as a framework for 'big data' integration, while avoiding overly extreme simplification, such as reducing various biological data types to the 'gene' level. A huge hurdle in developing such semantic knowledge models, or ontologies, is the knowledge acquisition bottleneck.Automated methods are still very limited and significant human expertise is required. In this research, we describe a methodology to systematize this knowledge acquisition and representation challenge, termed KNowledge Acquisition and Representation Methodology (KNARM). We also present how KNARM was applied on three ontologies - BioAssay Ontology (BAO), LINCS FramEwork Ontology (LIFE) and Drug Target Ontology (DTO) - built for three different projects - BioAssay Ontology, Library of Integrated Network-Based Cellular Signatures (LINCS) and Illuminating the Druggable Genome (IDG) -, and how they work together in complex queries. This is a Department of Computer Science PhD Proposal. Wednesday, 13th December 2017, 5:00pm, UB506 Dr. Paige Morgan will present Can We Make the Web Smarter? An Intro to Linked Open Data and the Semantic Web In 2001, Sir Tim Berners-Lee coined the phrase "semantic web" to describe an idea that he thought would revolutionize the way that the web worked: "a common framework allow[ing] data to be shared and reused across application, enterprise, and community boundaries." Since the idea of the semantic web was introduced, many people (both developers and non-developers) have made great progress towards realizing the vision of the semantic web - but we have also had to confront challenges involving technical infrastructure, maintenance, and authority. This talk will introduce the concepts of the semantic web, linked open data, and non-relational databases; and will examine specific current and potential applications, including Wikipedia, DBpedia, and a range of cultural heritage applications. Using these specific cases, the talk will also explore several questions and problems that the idea of the semantic web raises, and possible answers. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 6th December 2017, 5:00pm, UB506 Dr. Alessandro Warth will present Towards a Better Programming Experience - In the Classroom and Beyond Over the past few years, researchers have shared compelling visions of better, more helpful programming environments - e.g., my colleagues Bret Victor's "Inventing on Principle" (2012) and Sean McDirmid's "A Live Programming Experience" (2015) - but these visions have yet to impact the way we program today. The fact is there are substantial design and engineering challenges that prevent us from turning these researchers' compelling demos into practical and robust programming environments. My group at YC Research has been making progress on the Programming Experience (PX) problem. Our strategy is to first restrict the scope of the programming tools we create, then use what we've learned to branch out towards truly general programming environments. So far we have created a programming environment for implementing programming languages (the Ohm Editor), a new language that enables end-users to create mobile social apps (Chorus), and several other tools that aim to provide a better PX in the classroom, for teachers as well as students. This talk will focus on the latter: I will show some of the educational programming tools we have created, discuss their preliminary results, and outline our plans for next steps. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 29th November 2017, 5:00pm, UB506 Daniel Messinger Professor of Psychology University of Miami will present Sensing Typical and Atypical Development: What can Objective Measurement and Modeling of Interaction Tell Us? New sensing technologies and measurement algorithms are providing insight into the social and emotional development of typically developing children and those with communication disorders such as autism and deafness. Computer vision and pattern recognition tools have provided insights into emotional expression and the structure of early social interaction. Modeling of input from Kinect sensors is providing objective measurements of early attachment. Automated analysis of classroom movement and vocal interaction is suggesting how language development takes place among peers. The talk will consider the strengths, challenges, and future of objective measurement and modeling of child behavior. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 15th November 2017, 5:00pm, UB506 Amin Sarafraz Center for Computational Science University of Miami will present Photogrammetry: from Obscurity to Prominence The past few years have seen the emergence of exciting new techniques that produce 3D models from digital images. The underlying science for creating such models is called Photogrammetry and has been used for decades by surveying and civil engineers for mapping and documentation purposes. One of the main factors in unleashing the potentials of Photogrammetry and bringing it into prominence is the increase in computing power. I will give a brief overview of Photogrammetry and some of its applications fueled by ever increasing computing power. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 15th November 2017, 2:00pm, UB330G Mr. Joseph Masterjohn Department of Computer Science University of Miami will present Encasement: A Robust Method for Finding Intersections of Semi-algebraic Curves One of the fundamental concepts in computational geometry is deducing the combinatorial structure, or interactions, of a group of static geometric objects. In two dimensions, the objects in question include, but are not limited to: points, lines, line segments, polygons, and non-linear curves. There are various properties of interest describing a collection of such objects; examples include: distances, adjacencies, and most notably, intersections of these objects. Well studied, robust, and highly efficient algorithms exist for linear geometry and parametric curves. Problems involving non-linear, implicit, and high-dimensional objects however are an active area of research. Algebraic curves and algebraic surfaces arise frequently in numerous applications: GIS software, CAD software, VLSI design, computational chemistry and biology, dynamics, and robotics. We present a novel algorithm for finding all intersections of two semi-algebraic curves in a convex polygonal region, and describe its prospective analog in 3 dimensions. We "encase" the curves in the convex region by repeatedly splitting the region until each cell contains at most two intersecting segments, thus detecting and isolating all of the intersections. The advantage of using encasement is that the running time is proportional to the size of the convex region when it is small and yet comparable to existing techniques when it is large. This is a Department of Computer Science Masters Thesis defence. Friday, 3rd November 2017, 11:00am, Ungar 230 Dr. Francis J. Pelletier Department of Philosophy University of Albertta will present How to Make Some Many-Valued Logics be Useful In 1976 Nuel Belnap published a (relatively) well-known paper "A Useful Four-Valued Logic: How a Computer Should Think" in which he argued for a four-valued logic that could accommodate knowledge bases that were presented with conflicting data, or with incomplete data, about a query in some field of knowledge. More of the formal background was presented in Michael Dunn's "Intuitive Semantics for First Degree Entailment" that same year. The logic became known as FDE. A serious problem with FDE (and some related logics) is that you can't do any reasoning with them!! They do not contain a conditional connective that would allow for a normal modus ponens inference, nor from any other rule of inference that is equivalent to modus ponens, such as (unit) resolution and the like. The present talk proposes a conditional connective for FDE that will work, and furthermore has interesting (and surprising) extensions to certain 3-valued logics, such as Lukasiewicz-3, Kleene-3, RM3, and "the logic of paradox". This is joint work with Allen Hazen and (I hope) Geoff Sutcliffe. This is another in the Department of Computer Science Seminar series. Wednesday, 1st November 2017, 5:00pm, UB506 Dr. Adrian Reynolds University of Miami will present Learning How to Learn: Study Strategies Every Student Should Know Preparing our students to be innovators and leaders in their respective disciplines not only requires the development of subject matter expertise, but, equally important, the use of self-regulated learning strategies to solve complex problems of global significance. Regrettably, however, our education system has been structured around the teaching of content knowledge at the expense of teaching self-regulated active learning strategies that help to facilitate content mastery. Advising students to "take control of their own learning", as well-intentioned as it might be, might cause more harm than good if they're not taught how to do so. If you're a college student who can identify with any aspect of the following case, this seminar is for you: I consider myself a dedicated, hardworking student, but I somehow still struggle to perform to my full potential. When I study I mostly read the lecture slides and my notes multiple times so I can memorize key concepts. I sometimes study with a group and we quiz each other, but there's always that one person who's either behind or ahead of everyone else: This makes group work really difficult. I also create study guides which are more detailed, organized versions of my notes, but I rarely have time to go back and review all of them. I usually start reviewing a week before the test: If I begin reviewing any earlier, I'll forget the material. To stay organized, I usually keep a to-do list, but I hardly use a study schedule. I'm putting in the hours, but I feel like my grades don't reflect that. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 25th October 2017, 5:00pm, UB506 Dr. Dilip Sarkar Department of Computer Science University of Miami will present Threats and Vulnerabilities in Cyberspace Individual users, small businesses, large enterprises, governments, national and international infrastructures depend on cyberspace for regular operations. However, cyberspace and their users have vulnerabilities, which malicious actors are exploiting. In particular, malicious actors use /mal/icous soft/wares /(malwares), including viruses, worms, Trojan horses, backdoors, rootkits, adware, key loggers, spywares, ransomewares, and botnets. After discussing common methods used to distribute malwares, we will discuss bots and botnets, because they are the biggest threat in the cyberspace. In the past they have inflicted most harm to cyberspace infrastructures as well as cyberspace users; in the future they are expected to do so again and again. Also, we will discuss phishing and social engineering techniques. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 18th October 2017, 5:00pm, UB506 Dr. Alberto Cairo School of Communication University of Miami will present Visualization Design: It's Not Just About the Data When designing charts, graphs, or maps to inform the public, scientists often overlook critical features such as visual elegance, typography, color, hierarchy, and composition. They do it because they think that visual design is just about "beautifying" graphics. This talk will teach you to reconsider that thought. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 11th October 2017, 5:00pm, UB506 Dr. Timothy Norris Research Data Scientist University of Miami will present Participatory Research, Information Science and GIS Method Maps and geospatial information play an increasingly prominent role in the everyday lives of citizens in most "western" nations. Scholars also recognize that "the power of maps" stretches back in history to the origins of modern cartographic practice in the 15th and 16th centuries. Recent research shows that the nascent "data revolution" disrupts traditional power relations found in the visualization of geospatial information which can present opportunities for alternative mappings. One such alternative includes participatory mapping in which residents of a certain region or area become the cartographers who draw maps of their own territory. Some claim that participatory mapping can empower communities and increase democratic governance while others insist that the power of the map always lies with those who have the right to use force to legitimize the specific visualization of data. In this seminar, several participatory mapping projects will be presented and given the critique from a framework that includes critical cartography and on-the-ground experience with participatory mapping. The conclusion suggests that lessons learned from the participatory mapping experience can also be applied to allied data-driven good governance initiatives. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 4th October 2017, 5:00pm, UB506 Dr. Gang Ren Center for Computational Science University of Miami will present Content-Based Multimedia Timeline Analysis Employing Sequential Pattern Discovery and Alignment Multimedia timeline analysis explores the temporal patterns from multiple signal dimensions of multimedia content and employs these temporal patterns as a computational representation for modeling high-level multimedia concepts and objects, such as narrative structures, genres, emotion, and audience response attributes. An example of multimedia timeline analysis is the temporal arrangement and synchronization of the video color patterns and the audio event density in the content production processes, where these multimedia elements and their interactions render the narrative tension to enhance the storyline and persuasive effectiveness. We present a computational framework for multimedia timeline analysis based on sequential pattern discovery and alignment algorithms. This framework extracts video and audio signal features from the multimedia content and employs sequential pattern discovery tools to explore and interpret the temporal patterns in the signal feature timelines at hierarchical time resolutions. Our proposed framework enables the ensemble analyses of large multimedia datasets or more refined characterization of complex narrative dynamics, which are impossible or inefficient for conventional manual analysis workflows. Empirical studies and subjective rating experiments based on a commercial advertisement dataset are also included to demonstrate the usage of our proposed tools and to validate the cognitive relevance of the analysis results. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 27th September 2017, 5:00pm, UB506 Dr. Le Liu Department of Computer Science University of Miami will present Effective Visualizations of the Uncertainty in Hurricanes The National Hurricane Center track forecast cone is the most widely adopted visualization for enhancing viewers' understanding of hurricane forecasts. However, current research has experimentally shown that the cone easily leads to misconceptions of the uncertainty included, and difficulties of interpreting the crucial time-specific information. The research reported here focuses on mitigating these limitations by developing ensemble visualizations. The research began by developing a technique that generates and smoothly interpolates robust statistics from an ensemble of hurricane predictions, thus creating visualizations that inherently include the spatial uncertainty by displaying three levels of positional storm strike risk at a specific point in time. In addition, this research develops time-specific visualizations depicting spatial information based on a sampling technique that selects a small, representative subset from an ensemble of points. It also allows depictions of such important storm characteristics as size and intensity. Further, this research generalizes the representative sampling framework to process ensembles of forecast tracks, selecting a subset of tracks accurately preserving the original distributions of available storm characteristics and keeping appropriately defined spatial separations. This framework supports an additional hurricane visualization portraying prediction uncertainties implicitly by directly showing the members of the subset without the visual clutter. An ongoing cognitive studies suggests that these visualizations enhance viewers' ability to understand the predictions because they are potentially interpreted more as uncertainty distributions. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 6th September 2017, 5:00pm, UB506 Dr. Paige Morgan Digital Humanities Librarian University of Miami will present Can We Make the Web Smarter? An Intro to Linked Open Data and the Semantic Web In 2001, Sir Tim Berners-Lee coined the phrase "semantic web" to describe an idea that he thought would revolutionize the way that the web worked: "a common framework allow[ing] data to be shared and reused across application, enterprise, and community boundaries." Since the idea of the semantic web was introduced, many people (both developers and non-developers) have made great progress towards realizing the vision of the semantic web -- but we have also had to confront challenges involving technical infrastructure, maintenance, and authority. This talk will introduce the concepts of the semantic web, linked open data, and non-relational databases; and will examine specific current and potential applications, including Wikipedia, DBpedia, and a range of cultural heritage applications. Using these specific cases, the talk will also explore several questions and problems that the idea of the semantic web raises, and possible answers. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 30th August 2017, 5:00pm, UB506 Dr. Jonathan Poterjoy NOAA Atlantic Oceanographic and Atmospheric Laboratory will present Nonparametric Data Assimilation for Numerical Weather Prediction and Research Numerical weather prediction presents one of the largest applications of high-performance computing in science, owing to the enormous state dimension and the demand for timely forecasts. Weather forecasting also contains several intrinsic limitations related to the observability of atmospheric motions, nonlinearity in the underlying system dynamics, and the ability of numerical models to faithfully simulate the physical processes important for forecasting a given weather event. Therefore, a probabilistic framework is required. Quantifying probabilities for weather prediction is an exercise in Bayesian filtering, where a probability distribution for the model state is updated sequentially to reflect measurements collected from ground stations, aircrafts, satellites, and other sources. In geoscience, this process is called data assimilation. In this talk, I will introduce the basic strategies currently adopted for data assimilation in operational weather forecasting. These strategies rely on Gaussian assumptions for the underlying error distributions, which is often suboptimal when the model dynamics are nonlinear, when measurements relate nonlinearly to model state variables, and when observation errors are non-Gaussian. I will then introduce a new nonparametric data assimilation method that has potential for nonlinear/non-Gaussian problems encountered in weather models. Benefits of this method will be discussed using applications where convectively driven atmospheric motions determine the evolution of the dynamical system (i.e., severe thunderstorms and squall lines). This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 23rd August 2017, 5:00pm, UB506 Dr. Erik Noonburg Department of Biological Sciences Florida Atlantic University will present Modeling Animal Foraging Behavior as a Gtate-dependent Game Behavioral ecology often applies game theoretic models to understand animal decision-making. However, when decisions depend on an animal's state, analytic solutions for the game theoretic equilibrium are generally impossible. I will present two case studies in which I use reinforcement learning techniques to find approximate solutions. The case studies consider animals' choices of foraging locations in the face of competition, with application to wildlife management. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 26th April 2017, 5:00pm, UB230 Christopher Chung School of Architecture University of Miami will present RAD: Responsive Architecture + Design University of Miami RAD-UM provides resources and expertise for project-based research on the spatial ramifications of embedded technology and ubiquitous computing.The research is premised on the notion that every building or landscape component can be equipped with computational power. Projects at RAD-UM develop models for such digitally enhanced environments to better handle persistent and emerging challenges in the areas of healthcare, building technology and sustainability. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Friday, 24th April 2017, 5:00pm, Ungar 506 Dr. Nitesh Saxena University of Alabama at Birmingham will present All Your Voices Are Belong to Us: Stealing Voices to Fool Humans and Machines A person's voice is one of the most fundamental attributes that enables communication with others. However, equipped with the current advancement in automated speech synthesis, an attacker can build a very close model of a victim's voice after learning only a very limited number of samples in the victim's voice (e.g., mined through the Internet, or recorded via physical proximity). Specifically, the attacker can use voice morphing techniques to transform its voice-- speaking any arbitrary message-- into the victim's voice. In this talk, we will examine the aftermaths of such a voice impersonation capability, based on an off-the-shelf voice morphing tool, against three important applications and contexts: (1) impersonating the victim in a voice-based user authentication system, (2) mimicking the victim in arbitrary speech contexts (e.g., posting fake samples on the Internet or leaving fake voice messages), and (3) mimicking the victim in a Crypto Phone (e.g., Zfone, Silent Circle or Redphone) VoIP secure channel establishment process and thereby compromising the security and privacy of VoIP communications. When considering voice impersonation attacks against human users, we will also share our experiences running a neuro-imaging study (a part of a larger, "neuro-security" project), which provide root-level insights from neurological and cognitive perspective. This talk is based on a line of joint work with Maliheh Shirvanian, which appeared at the ACM CCS, ESORICS and ACSAC conferences in 2015, as well as a recent neuro-security study with Ajaya Neupane and other colleagues, currently being reviewed for publication. This is another in the Department of Computer Science Distinguished Scientist Seminar series. Wednesday, 19th April 2017, 5:00pm, UB230 Dr. Jie Xu Department of Electrical and Computer Engineering University of Miami will present Collaborative Mobile Edge Computing Pervasive mobile computing and the Internet of Things are driving the development of many new applications that are both compute-demanding and latency-sensitive. Although Cloud Computing enables convenient access to a centralized pool of configurable and powerful computing resources, it often cannot meet the stringent requirements of latency-sensitive applications due to the often unpredictable network latency and expensive bandwidth. The growing amount of distributed data further makes it impractical or resource-prohibitive to transport all the data over today's already-congested backbone networks to the remote cloud. As a remedy to these limitations, Edge Computing emerges as a new computing paradigm to push the frontier of computing applications, data, and services away from centralized cloud computing infrastructures to the logical extremes of a network thereby enabling analytics and knowledge generation to occur at the source of the data. However, many new challenges emerge and have to be addressed in this new computing paradigm to fully reap its benefit. In this talk, I will introduce some of our recent works on collaborative mobile edge computing. Game theoretic solutions are developed to enable efficient and secure collaboration among distributed edge devices in mobile networks, thereby significantly enhancing edge computing performance. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Friday, 14th April 2017, 2:30pm, Ungar 230 Dr. Jon Shlens Google will present A Learned Representation for Artistic Style The diversity of painting styles represents a rich visual vocabulary for the construction of an image. The degree to which one may learn and parsimoniously capture this visual vocabulary measures our understanding of the higher level features of paintings, if not images in general. In this work we investigate the construction of a single, scalable deep network that can parsimoniously capture the artistic style of a diversity of paintings. We demonstrate that such a network generalizes across a diversity of artistic styles by reducing a painting to a point in an embedding space. Importantly, this model permits a user to explore new painting styles by arbitrarily combining the styles learned from individual paintings. We hope that this work provides a useful step towards building rich models of paintings and offers a window on to the structure of the learned representation of artistic style. Jonathon Shlens received his Ph.D in computational neuroscience from UC San Diego in 2007 where his research focused on applying machine learning towards understanding visual processing in real biological systems. He was previously a research fellow at the Howard Hughes Medical Institute, a research engineer at Pixar Animation Studios and a Miller Fellow at UC Berkeley. He has been at Google Research since 2010 and is currently a research scientist focused on building scalable vision systems. During his time at Google, he has been a core contributor to deep learning systems including the recently open-sourced TensorFlow. His research interests have spanned the development of state-of-the-art image recognition systems and training algorithms for deep networks. This is another in the Department of Computer Science Distinguished Scientist Seminar series. Wednesday, 12th April 2017, 5:00pm, UB230 Dr. Mei-Ling Shyu Department of Electrical and Computer Engineering University of Miami will present Correlation-assisted Mining and Deep Learning for Semantic Concept Retrieval from Imbalanced Multimedia Big Data With the extensive use of smart devices and blooming popularity of social media websites such as Flickr, YouTube, Twitter, and Facebook, we have witnessed an explosion of multimedia data. As a result, multimedia semantic concept mining and retrieval whose objective is to mine useful information from the large amount of multimedia data including texts, images, and videos has become more important. Furthermore, many real-world datasets do not have uniform class distributions, and the minority class usually represents the concept of interests such as frauds in transactions, intrusions in network security, and unusual events in surveillance. The classifiers developed on datasets with such skewed distributions tend to favor the majority classes and are biased against the minority class. Despite extensive research efforts, the huge amount of multimedia data, the semantic gap issue, and the imbalanced concept retrieval remain challenging problems in multimedia research. To address these challenges, the joint efforts from big data, data mining, and deep learning for multimedia imbalanced data have been sought. In this talk, a novel correlation-assisted mining and deep learning system for semantic concept retrieval from imbalanced multimedia big data is introduced. To bridge the semantic gap, a concept correlation analysis model using the correlations between the retrieval scores and classes is proposed. To overcome the expensive computation issue, a convolutional neural network (CNN) based deep learning solution integrated with a bootstrapping technique is proposed. To handle big datasets, a Spark infrastructure has been built that shows promising performance of our proposed model with respect to feasibility and scalability. Finally, I will briefly introduce some of our other research work. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 5th April 2017, 5:00pm, UB230 Dr. Mark A. Finlayson School of Computing and Information Sciences Florida International University will present Automatically Extracting Narrative Structure: Results & Directions Narratives are a ubiquitous language phenomenon found in every society and culture. Perhaps the single most important feature that distinguishes narratives from other discourse is their structure: the structure of a narrative communicates its meaning and purpose and gives rise to numerous cognitive benefits that improve comprehension, retention, understanding, and use of information given in narrative form. I present results from the COGNAC Lab at FIU on automatically extracting narrative structure using approaches drawn from machine learning and computational linguistics. First, I present experiments that show that a classic theory of narrative structure (Vladimir Propp's morphology of the folktale) can be reliably reproduced by people. Second, I demonstrate a specially-designed learning algorithm can learn Propp's theory from raw data. Finally, I outline the most recent results from student work in the lab, including algorithms for story detection, character classification, and motif extraction, which point the way forward to systems for fully automatic narrative structure extraction and a major advance in machine intelligence. Dr. Mark Finlayson is Assistant Professor of Computer Science in the School of Computing and Information Sciences at Florida International University. He received his Ph.D. from MIT in 2011, and from 2012-2014 was a Research Scientist in MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL). This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Tuesday, 4th April 2017, 12:15pm, Ungar 330G Mr. Faisal Sikder Department of Computer Science University of Miami will present Human Fall and Activity Detection, and Muscle Spasm Identification Efficient image data compression algorithms are required to minimize the cost of data transmission and storage as the quality and file sizes of images keep increasing. With the advancements in image sensors and processing units of mobile devices, the use of complex but more effective compression algorithms is becoming more prevalent on a wide variety of devices. There are two types of compression algorithms for images: lossy (e.g. JPEG) and lossless (e.g. PNG). Lossy image compression results in information loss at the expense of compression gain. Lossless compression, in contrast, preserves image integrity fully, which is an important consideration in certain critical applications such as medical image processing. In this talk, we will first give an overview of the techniques and the applications of image compression. Then, we will present our research efforts on enhancing the performance of some of the existing algorithms and propose new ones to better gains of lossless image compression algorithms. More specifically, we will introduce pre-processing techniques (dynamic pseudo-distance matrix and scanning paths) and describe how we compress the pre-processed image data using the block-sorting transformations, and inversion coding technique along with an entropy coder. We will demonstrate, using various standard image data sets, that our proposed image compression techniques perform better than GIF, PNG, and JPEG 2000. Finally, we will discuss the implementation issues of our algorithms on microcontrollers (Arduino Uno, TI MSP432) as well as its parallelization on multi-processors. This is a Department of Computer Science PhD Defence. Friday, 31st March 2017, 2:30pm, Ungar 330G Mr. Basar Koc Department of Computer Science University of Miami will present Lossless Compression of Images Efficient image data compression algorithms are required to minimize the cost of data transmission and storage as the quality and file sizes of images keep increasing. With the advancements in image sensors and processing units of mobile devices, the use of complex but more effective compression algorithms is becoming more prevalent on a wide variety of devices. There are two types of compression algorithms for images: lossy (e.g. JPEG) and lossless (e.g. PNG). Lossy image compression results in information loss at the expense of compression gain. Lossless compression, in contrast, preserves image integrity fully, which is an important consideration in certain critical applications such as medical image processing. In this talk, we will first give an overview of the techniques and the applications of image compression. Then, we will present our research efforts on enhancing the performance of some of the existing algorithms and propose new ones to better gains of lossless image compression algorithms. More specifically, we will introduce pre-processing techniques (dynamic pseudo-distance matrix and scanning paths) and describe how we compress the pre-processed image data using the block-sorting transformations, and inversion coding technique along with an entropy coder. We will demonstrate, using various standard image data sets, that our proposed image compression techniques perform better than GIF, PNG, and JPEG 2000. Finally, we will discuss the implementation issues of our algorithms on microcontrollers (Arduino Uno, TI MSP432) as well as its parallelization on multi-processors. This is a Department of Computer Science PhD Defence. Wednesday, 29th March 2017, 5:00pm, UB230 Dr. Dilip Sarkar Department of Computer Science University of Miami will present Threats and Vulnerabilities in Cyberspace Individual users, small businesses, large enterprises, governments, national and international infrastructures depend on cyberspace for regular operations. However, cyberspace and their users have vulnerabilities, which malicious actors are exploiting. In particular, malicious actors use /mal/icous soft/wares/ (malwares), including viruses, worms, Trojan horses, backdoors, rootkits, adware, key loggers, spywares, ransomewares, and botnets. After discussing common methods used to distribute malwares, we will discuss bots and botnets, because they are the biggest threat in the cyberspace. In the past they have inflicted most harm to cyberspace infrastructures as well as cyberspace users; in the future they are expected to do so again and again. Also, we will discuss phishing and social engineering techniques. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 22nd March 2017, 5:00pm, UB230 Joel P. Zysman Center for Computational Science University of Miami will present The Internet of Things at University of Miami: Projects and resources available for researchers of all levels While we've all heard about the Internet of Things and how large it is going to become in the next few years; did you know that UM researchers are conducting IoT research in everything from Smart Cities to Climate Change? Please join Advanced Computing Director Joel Zysman for a round table discussion of both current research and the resources that are available for future work. See how UM researchers are using tools provided by the Center for Computational Science today and learn how to leverage these new technologies in your own work. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 8th March 2017, 5:00pm, UB230 Albert Harum-Alvarez SmallCo Lead Designer will present CODE SMELL: Why Tech Goes Stale Fast, and what that can mean for your career You're worried, and that's understandable. Some of the tech you're learning seems to be aging so fast that it will be obsolete before you can fully master it. While you're still young you're confident you can stay ahead of the curve, learning faster than the rate of obsolescence. But you read about middle-aged tech workers left behind, and you wonder if that could be you someday. The good news is that there are certain clear patterns in the way tech matures, and these patterns can be navigated and even managed. They are at the heart of app design at SmallCo [smallco.co]. SmallCo designs apps for clients such as Apple, Fidelity Investments, the investment bank BNP Paribas, the New York Public Library, Harvard University, The Cleveland Clinic and others. SmallCo's lead designer Albert Harum-Alvarez will give this talk, adapting material from his Design Masterclass, taught at sites around the world. SmallCo is looking for computer science students like you who want to study these concepts in depth, both to lengthen your tech career and because you really love to do good design. SmallCo is looking for talented interns willing to work and study at our offices in Miami, New York, Belgrade, Mexico City, Rotterdam and Goteborg Sweden. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 1st March 2017, 5:00pm, UB230 Mr. Basar Koc and Dr. Huseyin Kocak Department of Computer Science University of Miami will present Making of a website: www.perlforbiologists.org As an enhancement to the PRISM course CSC210 - Computing for Scientists, we have just unveiled the e-resource www.perlforbiologists.org consisting of a series of episodes featuring examples from genomics to help biologists learn the basics of the Perl programming language. After completing the final episode, students will be able to download a genome file from NCBI, and search and tally intricate motifs. The website design, simple yet elegant, is platform-independent and works well on mobile devices. The creation of the website is a purely educational undertaking and is devoid of any commercial material; free access is granted to all, not just to our students. In this talk, we will first discuss the choice of the content, the educational mission, and the design of the website. Then, we will introduce the following technical applications and standards that we used in the construction of the website: HTML5, Bootstrap, JavaScript, Google Analytics, Video Codecs (H264, H265, VP9), Podcast Audio Standards (ITU-R BS.1770-2), and Adobe Creative Cloud Apps. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 22nd February 2017, 5:00pm, MEA202 Faisal Sikder Department of Computer Science University of Miami will present Semi-Automatic Extraction of Training Examples from Sensor Readings for Fall and Activity Identification While inexpensive wearable motion-sensing devices have shown great promise for fall detection and human activity monitoring, two major challenges still exist and have to be solved: 1) a framework for the development of firmware, and 2) software to make intelligent decisions. We address both the problems. In this talk, we show our proposed generic framework for developing firmware. We also demonstrate that the k-means clustering algorithm can semi-automatically extract training examples from motion data. Moreover, we discuss about several one- and two level classification networks combinations of neural networks and softmax regression to monitor non-fall activities and to detect fall events. We also illustrate how the datasets for training and testing have been collected using the devices we assembled with four off-the-shelf components. This work advances the state-of the-art for development and training of wearable devices for monitoring non-fall activities and detecting fall events. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Friday, 17th February 2017, 2:30pm, Cox 42 Dr. Houssam Nassif Amazon will present Diversifying Amazon Recommendation Dr. Nassif will be presenting two recently published papers on diversifying Amaz on recommendations: "Diversifying Music Recommendations" (ICML'16 Workshop) uses submodular diversity to significantly improve Prime Music App recommendations quality and user engagement. "Adaptive, Personalized Diversity for Visual Discovery" (RecSys'16 Best Short Paper Award) describes Amazon Stream's seasonal, personalized and diversified recommendation framework. Amazon Stream (http://www.amazon.com/stream/), a new website for fashion discovery, uses Bayesian regression to score products, balances exploration and exploitation, applies submodularity to diversify recommendations, and learns seasonal and personalized weights to produce the final recommended personalized stream. This is another in the Department of Computer Science Seminar series, presented jointly with the Department of Neuroscience and the Center for Computational Science. Wednesday, 15th February 2017, 5:00pm, UB230 Dr. Athena Hadjixenofontos Center for Computational Science University of Miami will present The Many Forces of Nature: Lessons on the Genetic Architecture of Multiple Sclerosis from the Isolated Population of Sardinia Complex diseases are the most prevalent causes of death and disability in developed countries. Susceptibility to complex diseases is determined by the cumulative results of hundreds to thousands of genetic variants, as well as environmental exposures. In the last twenty years, numerous large scale projects have been undertaken to identify the genetic variants that underlie susceptibility to a number of complex diseases, including multiple sclerosis. Success has been limited: for multiple sclerosis, 110 genetic variants have been associated with the disease in outbred Caucasian populations, and it is estimated that hundreds to thousands more remain in the dark. The parameters that define the genetic architecture of a complex disease extend beyond the number of variants that underlie susceptibility, and include their effect sizes, population frequencies and whether or not they act additively. The limited success in forming a complete picture has led to the exploration of alternative hypotheses. Some of the alternative hypotheses target areas of the genetic architecture that traditional experiments are not designed to address. In this talk I will recount our contributions to understanding the genetic architecture of multiple sclerosis through the study of the isolated population of Sardinia. I will then lay out a set of future directions that are designed to challenge our assumptions about the hidden areas of genetic susceptibility, through forward-in-time genetic simulations. The completion of the proposed experiments will explain the reasons for our limited success in mapping disease variants and position us to choose the methods that can be effective in uncovering the remaining effects. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 8th February 2017, 5:00pm, UB230 Dr. Fabian Soto Department of Psychology Florida International University will present Extending Multidimensional Signal Detection Theory to Study the Independence of Brain Representation A common goal in visual neuroscience is to determine whether some stimulus dimensions (e.g., shape and spatial information, different shape properties, face expression and identity) are processed and represented independently from others. Such representation can be extremely useful; for example, if most objects can be recognized on the basis of a few shape dimensions, then representing those shape dimensions independently from any other visual information would allow fast object learning, by focusing attention only on the relevant dimensions and ignoring the irrelevant information. Furthermore, such learning would generalize broadly to any new object image from which the relevant shape dimensions can be extracted, regardless of how different this new image is from the training images. Such fast, generalizable learning is easily observed in people, but poorly understood. Unfortunately, analytical tools and models tailored specifically to study the independence of brain representations have not been developed. Without such tools, cognitive neuroscientists have resorted to proposing a multiplicity of operational tests of independence, each a small modification of traditional analyses adapted to measure a vaguely defined construct. Unsurprisingly, this research strategy has yielded contradictory results in most areas. This talk summarizes recent work aimed at solving this problem by extending General Recognition Theory, a multidimensional version of signal detection theory, to the study of independence of brain representations, including neuroimaging and neurophysiology. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 1st February 2017, 5:00pm, UB506 Dr. Geoff Sutcliffe Department of Computer Science University of Miami will present Automated Reasoning for the Dialetheic Logic RM3 A dialiethic logic allows formulae to be true, or false, or (differently from classical logic) both true and false, and the connectives are interpreted in terms of these three truth values. Consequently some inferences rules of classical logic are invalid in RM3, and some theorems of classical logic are not theorems of RM3. An automated theorem prover for RM3 has been developed, based on translations of RM3 formulae to classical first-order order logic, and use of an existing first-order theorem prover to reason over the translated formulae. Examples and results are provided to highlight the differences between reasoning in classical logic and in RM3. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Friday, 27th January 2017,2:30pm, Ungar 230 Dr. Hosna Jabbari Ingenuity Lab University of Alberta will present In Silico Tool to Improve Efficacy of Gene Therapy With the amount of genomic data produced every day, advances in medical sciences and development of new gene modification tools, a revolution in medicine is expected.Recent approval of the first gene therapies by FDA is a leap towards this revolution. Computational methods provide unique opportunities to realize this revolution by providing both an inexpensive framework (in terms of cost, time and safety) to explore the complex biological systems of diseases, and a reduced search space for treatment options. In this talk, I will describe an example of such framework for treatment of Duchenne muscular dystrophy through gene therapy, and highlight some of the challenges and future opportunities. This is another in the Department of Computer Science Seminar series. Wednesday, 25th January 2017, 5:00pm, Ungar 230 Dr. Tatiana Engel Department of Bioengineering Stanford University will present Discovering Dynamic Computations in the Brain from Large-Scale Neural Recordings Neuronal responses and behavior are influenced by internal brain states, such as arousal, vigilance, or task context. Ongoing variations of these internal states affect global patterns of neural activity, giving rise to apparent variability of neuronal responses to sensory stimuli, from trial to trial and across time within single trials. Demultiplexing these endogenously generated and externally driven signals proved difficult with traditional techniques based on trial-averaged responses of single neurons, which dismiss neural variability as noise. In this talk, I will describe my recent work leveraging multi-electrode neural activity recordings and computational models to uncover how internal brain states interact with goal-directed behavior. I will show that ensemble neural activity within single columns of the primate visual cortex spontaneously fluctuates between phases of vigorous (On) and faint (Off) spiking. These endogenous On-Off dynamics, reflecting global changes in arousal, are also modulated at a local scale during spatial attention and predict behavioral performance. I will also demonstrate that these On-Off dynamics provide a single unifying mechanism that explains general features of correlated variability classically observed in cortical responses (e.g., changes in neural correlations during attention). I will conclude by sketching out a roadmap for developing a general theory that will allow us to discover dynamic computations from large-scale neural recordings and to link these computations to behavior. This is another in the Department of Computer Science Seminar series. Monday, 23rd January 2017, 2:15pm, Ungar 230 Dr. Catie Chang National Institute of Health will present Uncovering New Dimensions of Human Brain Function from fMRI Data Functional magnetic resonance imaging (fMRI) is a powerful technique for human neuroscience. The richness and complexity of fMRI data present exciting challenges at the interface between computation and neuroscience, and require innovative data analysis methods together with deeper understanding of the neural and physiological basis of fMRI signals. I will describe my studies revealing features of brain function embedded in the dynamics of intrinsic brain networks. I will also discuss how, by integrating fMRI with electrophysiological, behavioral, and heart rate data, we uncovered components of fMRI dynamics related to vigilance and autonomic activity and developed a data-driven approach for detecting vigilance fluctuations in fMRI scans. These studies highlight ways in which previously unexplored dimensions of systems-level brain activity may be extracted from fMRI signals, and open new directions for neuroimaging biomarkers in health and disease. This is another in the Department of Computer Science Seminar series. Thursday, 19th January 2017, 8:30am, Ungar 230 Dr. Xuan Guo Oak Ridge National Laboratory will present Multi-omics Data Analyses via High-performance Computing for Complex Biological Systems Systems biology aims to model complex biological interactions at the system level by integrating information from interdisciplinary fields using a holistic perspective approach. Emerging high-throughput omics technologies promote current biology research into the age of systems biology and also raise huge challenges in multi-omics data integration, modeling, and systems-level analyses. High-performance computing techniques are promising to overcome the limits posed by conventional methods to the mining and exploration of large amounts of multi-omics data. In this talk, I will explore how to analyze high-throughput multi-omics data to better understand complex diseases and microbial communities. I will present several parallel algorithms and high-performance computing framework that are broadly applicable for the analyses of large data and complex biological systems. This is another in the Department of Computer Science Seminar series. Thursday, 12th January 2017, 8:30am, Ungar 230 Dr. Zheng Wang University of Southern Mississippi will present The Complex Systems of Protein Domain Co-Occurrences and the Three-Dimensional Structure of the Genome This presentation will start from a biological network named protein domain co-occurrence network that has not been fully studied but important, in which each node represents a protein domain, and if two domains co-exist in a protein, an edge is created to connect them. After repeating this procedure for all of the proteins of a species, a network of the species is created that contains species-specific biological signatures. Dr. Wang's research verifies that this type of network is scale-free network. Other topological properties including the shortest path distribution and the clustering coefficient distribution will also be shown in the presentation. A robustness test shows that this network is vulnerable to attack (remove nodes starting from the ones with the highest degree value) and robust to failure (remove nodes randomly). Dr. Wang has successfully used this type of network to predict protein and domain functions using neighbor-counting, biological function enrichment, and a machine learning algorithm based on the network topology. Another experiment of Dr. Wang has achieved an accuracy of 93.43% when applying this network to infer the phylogenetic relationships of 398 single-chromosome prokaryotic species using a graph alignment algorithm. After showing these research outcomes, the presentation will continue to a newly-emerged and potentially ground-breaking research topic, that is, the three-dimensional (3D) structure of the genome. The 3D conformations of healthy human chromosomes will be shown, followed by the comparisons of the intra-chromosomal spatial proximities of healthy, leukemia, and lymphoma human B-cells or cell-lines. The inter-chromosomal (between different chromosomes) spatial proximities and the chromosome translocation between chromosome 11 and 14 (a segment of chr.11 is exchanged with a segment of chr.14) in leukemia will also be illustrated in the presentation. Furthermore, a novel type of biological complex network will be introduced, which can indicate the 3D spatial proximities among biological components including protein coding genes, transcription factors, and lncRNAs. Using this type of novel complex network, Dr. Wang's lab has recently reconstructed the 3D structure of the chromosome X of mouse embryonic stem cells mapped with the localization intensities of Xist transcripts (an lncRNA that can inactivate the entire chromosome X by altering its 3D structure). Dr. Wang will present their algorithms and the corresponding reconstructed 3D structures in the resolutions of 500K base pair and, more excitingly, 40K base pair that has reached the gene level. After that, Dr. Wang's latest research in topological domains in mammalian genomes, which are recently believed to be the structural and functional units of the genome, will be briefly discussed. Particularly, Dr. Wang will show a novel structural measurement for topological domains and its correlations with genetic and epigenetic features. Together with Dr. Wang's internationally-recognized research in protein function prediction and protein model quality assessment using deep networks (stacked denoising autoencoders, SdAs), a newly-designed bioinformatics course proposed in a pending NSF CAREER proposal will also be briefly mentioned. Grant proposal ideas with computer science, physics, biological sciences, biostatistics, and medical school will be proposed throughout the presentation. This is another in the Department of Computer Science Seminar series. Monday, 5th December 2016, 2:30pm, UB230 Prof. Lance Fortnow Georgia Institute of Technology will present Bounding Rationality by Computational Complexity Traditional microeconomic theory treats individuals and institutions of completely understanding the consequences of their decisions given the information they have available. These assumptions may not be valid as we might have to solve hard computational problems to optimize our choices. What happens if we restrict the computational power of economic agents? There has been some work in economics treating computation as a fixed cost or simply considering the size of a program. This talk will explore a new direction bringing the rich tools of computational complexity into economic models, a tricky prospect where even basic concepts like "input size" are not well defined. We show how to incorporate computational complexity into a number of economic models including game theory, prediction markets, forecast testing, preference revelation and awareness. This talk will not assume any background in either economics or computational complexity. This is another in the Department of Computer Science Distinguished Scientist Seminar Series. Refreshments will be served immediately after the lecture in Ungar 330G. Wednesday, 30th November 2016, 5:00pm, UB230 Dr. Mahsa Mirzargar Department of Computer Science University of Miami will present Exploring the Potential of Data Depth for Uncertainty Characterization and Visualization of Ensembles When computational models or predictive simulations are used, researchers, analysts and decision makers are not only interested in understanding the data but also interested in understanding the uncertainty present in the data as well. In such situations, using ensembles is a common approach to account for the uncertainty, and explore the possible outcomes of a model. Visualization as an integral component of data-analysis task can significantly facilitate the communication of the characteristics of an ensemble including uncertainty information. In this talk, I will introduce novel ensemble visualization paradigms based on the generalization of conventional univariate boxplots and the concept of data depth. Generalizations of boxplot provide an intuitive yet rigorous approach to studying variability and descriptive features of an ensemble. The nonparametric nature of this type of analysis makes it an advantageous approach to study uncertainty in various applications ranging from image analysis to fluid simulation to weather and climate modeling. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 16th November 2016, 5:00pm, UB230 Dr. Nurcin Celik Department of Industrial Engineering University of Miami will present Data-and Computation-aware Modeling for Resilient Power Grids must view the real system as a dynamic entity and coordinate decisions at different levels. This requires continual and timely updating of its sensors, communications, and the models that support its control. In this research, our goal is to investigate a novel adaptive simulations architecture, namely Dynamic Data Driven Adaptive Multi- scale Simulations (DDDAMS), to obtain accurate system state estimations and initiate reliable operational tasks for microgrids by integrating dynamic data into the simulation in a timely manner through automatic fidelity switching. Proposed framework includes 1) real-time adaptive simulation, 2) grid computing modules (for computational resource management), 3) web services (for communications), 4) database, 5) various sensors, 6) real system, and 7) algorithms for state estimation, fidelity selection and assignment, and task generation involving Bayesian inferencing. The microgrid considered in this study embodies sources for both conventional and renewable energy generation as well as its side necessity of storage capacities including diesel generators, and test sites for solar, wind, and ocean energy. The latter two have become equally important nowadays due to the increasing growth in energy demand, insufficiency of natural resources, and newly established policies for low carbon footprint. A prototype of the proposed framework is built for a medium scale electric microgrid with a capability to isolate from the main utility with its dedicated substation and feeders. While being developed, the components of the proposed framework have been demonstrated for the economic load dispatch problem in microgrids using systematic formal methods, network simulation with collected sensory data, and expertise from subject matter experts. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 9th November 2016, 5:00pm, UB230 Dr. Suhrud Rajguru Department of Biomedical Engineering and Otolaryngology University of Miami will present Modeling Inner Ear Mechanics for Diagnostic and Therapeutic Applications In this seminar, I will present results from studies focused on modeling Benign Paroxysmal Positional Vertigo (BPPV), one of the most commonly experienced vertigo syndrome. We have developed mathematical and experimental models of various BPPV mechanisms. I will discuss engineering, experimental, and modeling perspectives. Results presented have improved the understanding, assessment and treatment of variant forms of this complex inner ear syndrome. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 2nd November 2016, 5:00pm, UB230 Dr. Geoff Sutcliffe Department of Computer Science University of Miami will present Hoping for the Truth - A Survey of the TPTP Logics This work compares features of the classical logics that are commonly used in the TPTP-based automated reasoning community for representing chosen aspects of "the world", and the consequent implications for reasoning about these representations. It is argued that increases in complexity in terms of representation and reasoning force users to compromise between the reliability of the representation and the reliability of the reasoning. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 26th October 2016, 5:00pm, UB506 Mr. Nasir Uddin Laskar Department of Computer Science University of Miami will present What is Deep Learning and How is it Useful for Understanding the Visual Brain? In recent years, there has been increasing interest in an area of machine learning known as Deep Learning. These techniques have become state of the art in visual object and speech recognition and are now making significant progress in robotics, language translation, bioinformatics, and many more fields. Deep learning borrows inspiration from the hierarchical organization observed in the brain, such as the cortical visual processing of objects across multiple layers. But how visual information is represented in the layered structure of the brain is not well understood beyond the primary visual cortex (first layer). Utilizing advances in deep learning, we show there is a good correspondence across a range of metrics between the second layer of a deep convolutional neural network and recent biological experiments in secondary visual cortex revealing selectivity of the neural population to textures. Our findings will accelerate the understanding of mid-level visual cortical representations in the brain and how they develop hierarchically. Dr. Luis Gonzalo Sánchez Giraldo Department of Computer Science University of Miami will present Integrating Nonlinearities into Deep Neural Networks Deep learning for image classification has benefited from incorporating nonlinearities inspired by Neuroscience, such as divisive normalization, which is an ubiquitous computation in neural processing. However, in practice, nonlinearities are usually chosen in a fixed and somewhat ad hoc manner and not learned from the data. We propose that the brain has adopted?and that artificial systems can benefit from?a more optimal strategy for choosing nonlinearities based on the statistical properties of its inputs. We develop an approach that integrates more realistic neural nonlinearities previously used to model the first layer of visual cortex, into deep networks. This is expected to shed light on the types of nonlinearities employed in middle cortex and also lead to improved technologies for artificial systems. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 19th October 2016, 5:00pm, UB230 Dr. Milica Mormann School of Business Administration University of Miami will present The Role of Visual Salience and Attention in Economic Decision-Making Many decisions we make require visually identifying and evaluating numerous alternatives quickly. These usually vary in reward, or value (e.g., how much you like a box of cookies), and in low-level visual properties, such as saliency (e.g., how visually salient the cookie box is). Both saliency and value influence the final decision. In particular, saliency affects fixation locations and durations, which are predictive of choices. I examine the relationship between bottom-up saliency and top-down reward and their role in economic decision making, using both behavioral experiments and computational modeling approaches. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 12th October 2016, 5:00pm, UB506 Dr. Sheyum Syed Department of Physics University of Miami will present The Fruit fly As a Model to Elucidate the Emergence of Behavior Can a machine with ~1011 parts be understood in terms of a "simpler" one with only 105 parts? The answer to this question is an optimistic and measured "yes". Like the enduring toy problems in physics and computer science, model organisms such as the fruit fly and the mouse have for long served as testbeds in biology. My laboratory uses the fruit fly with its 105 neurons to try to make sense of the basic principles presumably at play in the human brain which has approximately 1011 neurons. In this talk, I will describe our approach to delineating neural principles by combining an assortment of ideas and methods from genetics, physics and mathematics. I will show examples of video imaging of behaving flies, computational interpretations of behaviors, targeted genetic perturbations of behavioral circuits and lastly, construction of models with simple predictive power. Along the way I will provide a parochial view of areas where computer science has and will continue to play critical role in our collective effort towards understanding the brain. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 28th September 2016, 5:00pm, UB230 Dr. Ben Kirtman Department of Atmospheric Sciences University of Miami will present Climate Change Science and the Computational Challenge This talk describes the scientific basis for our understanding of climate change, and how computer models are used to assess how human activities affect climate. A brief description of the state-of-the-art climate modeling is provided. In this aspect of the talk the focus is on critical processes that are missing, and affect our ability to predict regional change. These missing processes are intimately connected to computational limitations. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 21st September 2016, 5:00pm, UB230 Dr. Chaoming Song Department of Physics University of Miami will present Quantifying Group Dynamics in Online Social Media The emergent processes driving social dynamics are a product of complex interactions among large numbers of individuals. In this talk, I will show several examples of group dynamics across various online social media. We will focus on sudden uprisings occurred in social unrests, providing evidence instead of a remarkable gestational phase marked by self-organized aggregation through Social Networking Sites. Development hyper-escalates ahead of an uprising, enabling prediction of the real-world onset with substantial lead-times. Following a close parallel with the correlation clustering in quasi-particles and cyber-enabled contentious politics, we develop a theory of multi-agent adaptation agnostic of country and language. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 14th September 2016, 5:00pm, MEA 202 Heather Ciechowski and Scott Joseph Google will present Applied Computer Science & Engineering at Google Why does this matter? Will I ever actually use this stuff? Ever wonder how the concepts you?re learning in your computer science and engineering courses pertain to the ?real world?? Join two Google engineers/UM alumni as they discuss how these concepts are used in real Google products, such as YouTube and Maps, as well as in their own work. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 7th September 2016, 5:00pm, Cox Neuroscience Annex 120 Dr. Dan Isom Department of Molecular and Cellular Pharmacology University of Miami will present Analytical Mining of 3-Dimensional Protein Structures Using Computational Geometry Proteins are nature's nano machines. From harnessing the energy of the sun, to coordinating the spontaneous self-assembly of biological structures, proteins are responsible for every emergent property of life. A protein's function is determined by its 3-dimensional shape - proteins spontaneously "fold" into specific shapes as they are synthesized. In my talk I will discuss analytics-based techniques that my lab is developing to better understand how a protein's shape is related to its biological function. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Wednesday, 31st August 2016, 5:00pm, Ungar 230 Dr. Daniel Messinger Department of Psychology University of Miami will present Emotions, Interaction, and Autism: Leveraging Big Behavioral Data Infants develop in the context of social interaction with parents and others. Study of these social interactions typically involves behavioral observation using expert coding systems that rely on subjective judgements. To produce objective measurements of infant and parent behavior, my lab employs computer vision and pattern recognition approaches. These methods produce "big behavioral data" fine-grained records of expression and other movements in time. To make sense of the underlying interactive process in these observations, we employ computational approaches such as inverse optimal control and statistical approaches such as dynamic time-series models. The ultimate goal is understanding and predicting both typical development and developmental disturbances such as autism spectrum disorder (ASD). To this end, we incorporate common genetic (e.g., dopaminergic) variants in our models of interaction and development. This is another in the Department of Computer Science Pizza Seminar Series. Refreshments will be served at 4:30 p.m. in the reception area of the 3rd floor of Ungar. Monday, 22nd August 2016, 3:00pm, Ungar 330G Ms Parul Maheshwari Department of Computer Science University of Miami will present A Model-based Sensor Database for the Internet of Things The Internet of Things (IoTs) is becoming ubiquitous in our everyday lives, hence more technologies will generate data. Sensors in IoT devices monitor various attributes of the environment and produce data periodically and storing this massive data in the database is becoming a huge challenge in the data storage infrastructure. Prior research has proposed compression algorithms and signature techniques to reduce data storage but do not specify how the data patterns are defined. In this study, we propose a system that stores data models instead of raw data points. This helps in reducing data storage requirements. The data models developed are mathematical polynomial models that fit a sample data set. In addition, we propose a sensor database structure that addresses the issues of data redundancy as well as temporal constraints in the database. This is a Department of Computer Science Master's Thesis Defense. Wednesday, 20th April 2016, 5:00pm, UB230 Dr. Martin Golumbic Professor of Computer Science and Director of the Caesarea Rothschild Institute for Interdisciplinary Applications of Computer Science University of Haifa will present Graph Theoretic Perspectives on Reasoning about Time Reasoning and acting within the time constraints of the real world are among the most fundamental notions of intelligence. Understanding the nature and structure of such constraints can help to find a satisfying solution or find a relaxation when no solution can be found. Given certain explicit temporal relationships between events, we may have the ability to infer additional relationships which are implicit in those given. For example, the transitivity of "before" and "contains" may allow inferring information regarding the sequence of events. Such inferences are essential in story understanding, planning and causal reasoning. Temporal information may be qualitative where events are represented by abstract time points and time intervals, and we process and deduce relationships between them, such as pairs intersecting each other, one preceding, following or containing another, etc. Other information may be quantitative where durations can be measured, precise time stamps may be available, or numerical methods can be applied to understand a specific time line of events. We will explore a variety of these topics with an emphasis on graph theoretic models and algorithms. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 13th April 2016, 5:00pm, UB230 Dr. Manohar Murthi Department of Electrical and Computer Engineering University of Miami will present Signal Processing for Engineering and Understanding Networks with a Focus on Mobile Adaptive Networks Signal processing methods are increasingly being used in scenarios involving graphs or networks. For example, signal processing methods can be used to engineer the design of a network of mobile agents that exploit distributed learning and adaptation to achieve a task. Signal processing methods are also used to understand phenomena in natural networks, such as the spread of opinions in social networks. In this talk we will give a brief overview of some work on these topics, focusing on the design and performance analysis of an algorithm designed for mobile agents pursuing multiple stationary targets. This algorithm allows mobile agents to split up and form distinct clusters to learn about and pursue multiple targets, all while moving in a cohesive collision free manner. (There will be videos!). Time permitting, we will also talk about some work on estimating latent attitudes in social networks based on limited observations. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 6th April 2016, 5:00pm, UB230 Dr. Abhishek Prasad Department of Biomedical Engineering University of Miami will present Hybrid Computational and Biologic Decoding for Brain Machine Interfaces Neuroprosthetics can provide partial restoration of motor function in people living with paralysis. The neuroprosthetic controller, which translates user's motor intent into the prosthetic actions, should ideally be able to adapt and learn the optimal control policy autonomously, using, for example, feedback from the user. Hebbian reinforcement learning (HRL) in a connectionist network provides an adaptive controller design that learns using only a binary evaluative feedback signals, as a measure of desirability/undesirability of performance. We developed an alternative brain machine interface (BMI) framework based on an actor-critic Reinforcement Learning (RL) paradigm that eliminates the need for a supervised error signal. The Actor component of the Actor-Critic RL is driven by the primary motor cortex (MI) which decides on an action based on a fully connected neural network and the Critic component is driven by the nucleus accumbens (NAcc), which gives an evaluative feedback on the action taken. This method, which combines the efficiency of supervised learning with the generality of reinforcement learning, offers a promising approach to facilitate the usage of neuroprosthetic systems as practical tools in daily-life activities. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 30th March 2016, 5:00pm, UB230 Gavin Heale Kugadi will present A Startup Primer Startup activity in the USA rose rapidly in 2015 experiencing the largest year-over-year increasefor the past two decades. Ever wonder what it's like to join that trend either as a new owner or as an employee? We'll be taking a quick tour of how startups are formed, funded and produce new products. Along the way, we'll discuss practical tips to understanding the startup economy as well as a look at the rapidly growing Miami startup ecosystem. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 23rd March 2016, 5:00pm, MAE 220 Prof Shahriar Negahdaripour Department of Electrical and Computer Engineering University of Miami will present 2-D sonar Image Utilization in the 3-D Interpretation of World Objects Machine vision systems have provided numerous key capabilities in the deployment of autonomous terrestrial robotics systems, and submersibles when operating in relatively clear waters. However, these become ineffective when subsea robotics systems are deployed in turbid environments. Under these circumstances, realizing capabilities similar to those in clear water is highly desirable. This talk will cover a range of methods for the automated processing the FS sonar video data to extract various 3-D information, aimed at enabling automated operation of subsea robotics platforms and (or) reconstruction of imaged objects and environments. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 16th March 2016, 5:00pm, Ungar 230 Heather Ciechowski Google will present Life of a Software Engineer Have you ever wondered what it's actually like to be a software engineer? You can probably imagine part of the job is writing code, but is that it? What else do they do? How much do they interact with people? What tools do they use? What skills do they need that they didn't learn in college? Join us as we answer these questions and discuss the day-to-day life of a software engineer. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 2nd March 2016, 5:00pm, Ungar 230 Dr. Lucina Q. Uddin Department of Psychology University of Miami will present Human Connectomics: Imaging and Analysis of Large-scale Brain Networks Brain structural and functional development underlies the maturation of increasingly sophisticated cognitive abilities. High-level attentional and cognitive control processes rely on the integrity of, and dynamic interactions between, several core neurocognitive networks. The right fronto-insular cortex (rFIC) is a critical component of a salience network that is thought to mediate interactions between large-scale brain networks involved in externally oriented attention and internally oriented cognition. How these systems reconfigure with development is a critical question for cognitive neuroscience, with implications for neurodevelopmental pathologies affecting brain connectivity. I will describe studies examining interactions within and between these systems in typical and atypical development. Additionally, I will present results of novel dynamic functional connectivity analyses of subregions within the rFIC and show that functional connectivity-based classification can be used to successfully discriminate children with autism from their typically developing peers. These findings from adults, typically developing children, and children with autism suggest that structural and functional maturation of rFIC pathways is a critical component of the process by which human brain networks mature to support complex, flexible cognitive processes in adulthood. In describing these empirical findings, I will introduce the most commonly used methods to study the development of brain connectivity. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 24th February 2016, 5:00pm, Ungar 230 Jose Perez Time4Learning will present Overcoming Agile Methodology Pitfalls Since its introduction in 2001, the agile methodology has been replacing the waterfall methodology with an iterative and dynamic process capable of adjusting to the new demands for flexibility and rapid change in the software development industry. Many adopters of the methodology describe it as a remedy to most problems we face when we develop software. Nevertheless the methodology has several pitfalls that must be overcome if a company is to be successful in its use. The cost of ignoring the agile methodology pitfalls could set the software development team and managers on a path to cyclical and continuous failure. In this seminar, the speaker will describe how to identify and address the root causes of these pitfalls from a software engineering management perspective. After the talk, the speaker will answer questions from students regarding internships at Time4Learning. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 17th February 2016, 5:00pm, Ungar 230 Dr. Tarek Sayed Department of Business Technology University of Miami will present Emerging Technology Themes This lecture will focus on emerging technology themes (IoT, Big Data and Robotics) and their transformative impact on business and law. The lecture will focus primarily on business applications and some of the legal issues and different legislative attempts to regulate these emerging technologies - the good and the bad. In addition to exploring the fast growing adoptions of Big Data, IoT and Robotics by a multitude of industries and organizations, we will discuss issues related to privacy, liabilities, and fascinating aspects of a forward looking modeling of some legal frameworks prompted by advances and adoptions of these technologies. As time permits, we will also explore some of the recent business and legal cases with respect to Big Data, rights to data, and privacy. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 10th February 2016, 5:00pm, Ungar 230 Dr. Dilip Sarkar Department of Computer Science University of Miami will present Sensor Data Aggregation in Presence of Malicious Attacks Phenomenal advancement of inexpensive sensors and microcontroller technologies have significantly contributed to development of Internet of Everythings (IoEs). Most IoE devices have multiple types of sensors embedded in them. These sensors are unreliable and vulnerable to attacks, including malicious attacks. To overcome unreliability of readings from a sensor, readings from multiple sensors are gathered by an aggregator node. In this talk we discuss a method for detecting reliable sensors. Once reliable sensors are identified, we use their readings for true signal value estimation. A sensor node or IoT device is considered good (bad), if all sensors embedded in it are reliable (unreliable). We also discuss a method for identification of good (bad) sensor-nodes. Extensive evaluations of the proposed algorithms show that they identify good and bad nodes, and estimate true sensor value efficiently. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 3rd February 2016, 5:00pm, Ungar 230 Renzo Silva and Charlie Saravia Fintech will present FinTech Start-Up Overview: P1 Analytics We will provide an overview of the financial tech industry and how it is disrupting established players. A case study of the Miami-based funded FinTech start-up, P1 Analytics, will be presented by its co-founders. P1 Analytics develops cutting-edge investment software and analytics for retail and professional users. Technologies used include IOS, Android, MySQL, PHP, C/C++, and JQuery. The presentation will include a demo of their beta social investing app, Stocli. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Monday, 1st February 2016, 11:00am, Ungar 230 Dr. BC Kwon IBM Research NY will present Designing Human-Centered Visualizations for Data Analysis Data visualization, combined with data mining algorithms, is a powerful way for users to explore vast amounts of data and to communicate findings. My research aims to investigate visual analytics methods tailored to human's capabilities and needs. Human's capabilities vary depending on their expertise, interest, visualization literacy, and others. Before we provide visualization solutions to domain problems, we need to identify users' task goals and requirements as well as users' limitations. I have pursued my research goal by conducting empirical studies as well as designing novel visual analytic methods with experts and non-experts from various domains, from investigative analysis to health care. In this talk, I introduce my previous work in empirical studies as well as application studies. In particular, I introduce my most recent visualization system design study, called VisOHC, which is a visual analytics system for online health community administrators. At the end, I provide several takeaway lessons and plans for ongoing and future studies. This is another in the Department of Computer Science Seminar series. Wednesday, 27th January 2016, 5:00pm, Ungar 230 Dr. Neil Johnson Department of Physics University of Miami will present Cyber-Physical Networks and the Race to Zero Cyberphysical systems (CPS) are set to become ever faster, driven by computational and technological advances that push them toward speed limits set by fundamental physics. Application areas in everyday life include market exchange systems, resource-allocation systems and remote sensing systems. As highlighted recently by the NSF, there is need for a theory of the dynamical behavior of CPS systems, that allows for networking at multiple scales, coupling across multiple temporal and spatial scales, imperfect network communications and sensors, as well as adaptive reorganization and reconfiguration of the system. This talk presents initial progress on a project toward this goal. Specifically, I will present a computationally-inspired toy model of the dynamics in decentralized networks of semi-autonomous machines, in which an ecology of algorithms, sensors and network links may be operating, adapting and even competing in response to external inputs. Attention is paid to the regime of sub-second behavior where human intervention and hence real-time management becomes impossible - and where the system is almost entirely self-organized. The availability of data from the network of US financial exchanges provides a test-bed for the analyses being developed. Ultrafast anomalies have already been observed in such CPS systems. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Monday, 25th January 2016, 11:00am, Ungar 230 Dr. Nivan Ferreira University of Arizona will present Interactive Visualization Techniques for Urban Data Analysis The explosion in the volume of data about urban environments has opened up opportunities to better inform both policy and administration, and thereby to help governments to overcome constant challenges of improving/increasing the quality of public services and promoting sustainable development. By taking advantage of the modern computer graphics and the power of the human visual system, interactive visualization techniques have been shown to be powerful tools that help making sense of large collections of data. In this talk, I will describe recent data visualization techniques designed to allow analysts to interactively explore and analyze large collections of urban data. These techniques include visual and algorithmic aspects and have been applied to help domain experts in the fields of urban planning, transportation engineering, and architecture. This is another in the Department of Computer Science Seminar series. Wednesday, 20th January 2016, 5:00pm, Ungar 230 Dr. Alberto Cairo School of Communication University of Miami will present A Truthful Art: Charts and Visualization for Effective Communication In the past two decades, visualization has become a language adopted not just by scientists and statisticians, but by journalists, designers, and the public in general. This process has been enabled by an increasing variety of software tools that are sophisticated, easy to use and, in many cases, free. However, knowledge of elementary principles of information presentation has not progressed at the same pace. This talk will provide an overview of these principles. You will learn: • How to create charts and maps that don't mislead • How to choose appropriate graphic forms to represent your information • How to arrange your graphics to build compelling narratives This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served at 4:30 pm in the reception area of the 3rd floor of Ungar. Wednesday, 20th January 2016, 11:00am, Ungar 230 Dr. Mahsa Mirzargar SCI Institute, University of Utah will present Revisiting Data Analysis through Scientific Visualization Computational and data-intensive methods are transforming scientific discovery. The complexity and heterogeneity of scientific data is revealing the limitations of the current data analysis techniques to support reasoning and the decision-making process in today's multidisciplinary problems. Visualization, as an integral component of data analysis task, can significantly facilitate the exploration, characterization and communication of the data. In this talk, I will present examples in which researchers, analysts and decision makers are not only interested in understanding the data, but also interested in understanding the uncertainty present in the data. I will explore ideas within the context of uncertainty characterization and visualization that are broadly applicable for the analysis of complex and heterogeneous data. This is another in the Department of Computer Science Seminar series. Thursday, 10th December 2015, 11:30am, Ungar 230 Tao Wang University of Waterloo will present Attacks and Defenses for Website Fingerprinting Website fingerprinting attacks allow a local, passive eavesdropper to determine a client's web activity by leveraging features from her packet sequence. These attacks break the privacy expected by users of privacy technologies, including low-latency anonymity networks such as proxies, VPNs or Tor. They are invisible to the victim and difficult to defend against. As a discipline, website fingerprinting is an application of machine learning techniques to the diverse field of privacy. In this talk, I will describe my new algorithms for both website fingerprinting attacks and defenses. I will show that website fingerprinting poses a realistic threat to internet users, and demonstrate that my attacks are successful in breaking the current defenses on Tor. I will then give a construction of the first website fingerprinting defense that is both effective and efficient. This is another in the Department of Computer Science Seminar series. Monday, 7th December 2015, 11:30am, Ungar 230 Yanmin Gong University of Florida will present Privacy-Preserving Data Analytics for Big Data Applications Big data has driven innovation, productivity, efficiency, and growth in many domains, creating enormous benefits for the global economy. However, with massive data and advanced data analytical techniques, far more information can be inferred than most people have anticipated at the time of data collection/publication. Traditional privacy-preserving techniques are either insufficient against such new privacy attacks or preventing reasonable data usage. We need a secure and privacy-preserving solution which allows people to learn information as it was intended and stops people from learning information in ways it was not. In this seminar, I will first describe privacy challenges in current big data applications , and then focus on protecting the privacy of biomedical "big" data in healthcare systems. Our work attempts to design a practical, secure, and privacy-preserving framework for utilizing healthcare data which are distributed among multiple parties. Our framework is general and can be used for multiple learning algorithms. Moreover, I will briefly introduce my work on security and privacy in other big data applications such as the mobile crowdsourcing and smart grids. This is another in the Department of Computer Science Seminar series. Wednesday, 2nd December 2015, 5:00pm, Ungar 506 Dr. Mitsunori Ogihara Department of Computer Science University of Miami will present Data Mining for Non-scientific Research Data mining is a research that aims at developing techniques for discovering interesting patterns in datasets that are otherwise trashed. The use of data mining for knowledge discovery was limited to businesses, but now, thanks to the proliferation of high-speed Internet connections and data storage, data mining has been creeping into other areas, such as biology and medicine. In this talk I will present and discuss some instances of data mining for non-scientific disciplines, such as humanities and music. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Tuesday, 1st December 2015, 11:30am, Ungar 230 Dr. Shouling Ji Georgia Institute of Technology will present Evaluating the Security of Anonymized Big Graph/Structural Data Nowadays, many computer systems generate structured data (also called graph data). Graph data spans many different domains, ranging from online social network data from networks like Facebook to epidemiological data used to study the spread of infectious diseases. Graph data is shared regularly for many purposes including academic research and for business collaborations. Since graph data may be sensitive, data owners often use various anonymization techniques that often compromise the resulting utility of the anonymized data. To make matters worse, there are several state-of-the-art structured data de-anonymization attacks that have proven successful in recent years. To date, graph data owners cannot gauge the practical or theoretical vulnerability of their data, nor can they comprehensively gauge its utility after anonymization. In this talk, we first introduce various novel structure-based de-anonymization attacks on graph data. Subsequently, we study the theoretical foundation for the success of existing de-anonymization attacks along with large-scale evaluations on real-world graph data. Third, we propose, design, and implement SecGraph, a uniform and open-source Secure Graph data sharing/publishing system. Finally, we will discuss some future research directions. This is another in the Department of Computer Science Seminar series. Wednesday, 11th November 2015, 5:00pm, Ungar 506 Dr. Daisy Zhe Wang Department of Computer and Information Science and Engineering University of Florida will present Data Science Research In this talk, I discuss novel system components and algorithms that we are designing and building at UF to enable a probabilistic master Knowledge Base (KB) system. In the context of the Archimedes project, I will discuss a spectrum of research directions we are exploring at the UF Data Science Research (DSR) group including: query-driven and scalable statistical inference, probabilistic data models, state-parallel and data parallel data analytics framework, multimodal (e.g., text, image) information extraction, and KB schema enrichment. This line of research of supporting analytics over automatically extracted knowledge bases is of high impact for many applications from QA systems, situational awareness to medical informatics. Other related projects include DeepDive from Stanford, YAGO from Max Planck Institute, NELL from CMU as well as WikiData/Freebase and Google Knowledge Vault. Apart from research, I am also happy to discuss Data Science courses at different levels and cross disciplinary education. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Wednesday, 4th November 2015, 5:00pm, Ungar 506 Dr. Elisha Sacks Purdue University will present Solid Modelling for VLSI Design Solid modelling is a core technology in VLSI process and device modelling. Intel has found that commercial software can be unreliable and slow. An important example is computing an offset of a polyhedron. Prior work on robust computational geometry makes assumptions that conflict with the application requirements, notably the need to perform finite-element analysis on solid models. I will describe preliminary work on a robustness technique that meets these requirements. I will present a novel approximate offset algorithm that I have implemented using this robustness technique. I will conclude with a discussion of the outstanding problem of eliminating small features from polyhedral meshes. Joint work with Dr. Victor Milenkovic. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Tuesday, 3rd November 2015, 11:30am, Ungar 330G Mr. Anes Yessembayev Department of Computer Science University of Miami will present Secure Data Aggregation for Sensor Networks in the Presence of Collusion Attack using Local Outlier Factor Aggregation of data from multiple sensor nodes is usually done by simple methods such as averaging or, more sophisticated, iterative filtering methods. However, such aggregation methods are highly vulnerable to malicious attacks. In this work, algorithms that eliminate or minimize influence of false readings are developed and evaluated. This is a Department of Computer Science Master's Thesis Defense. Wednesday, October 28th 2015, 5:00pm, Ungar 506 Dr. Mihai Polceanu Florida International University will present ORPHEUS: Reasoning and Prediction with Heterogeneous Representations Using Simulation Interactive virtual environments pose a wide variety of challenges for intelligent agents, especially to make decisions in order to reach their goals. The difficulty of decision making tasks rises quickly by introducing continuous space and real time into question. It also becomes increasingly harder to build intelligent agents that can meaningfully interpret and act in unknown situations. We take inspiration from cognitive science, specifically from how humans perform mental simulation to anticipate events in the world around them, with the aim of obtaining an autonomous agent that makes decisions and adapts itself to novel situations. The mental simulation paradigm enjoys significant interest from the cognitive science community, but computational approaches to mental simulation rely on specialized simulators for a given task and thus are limited to specific scenarios. Our contribution is a generic agent architecture (ORPHEUS) which supports decision-making based on the simulation of functional models of the world ahead of time, inspired from how humans imagine the outer world and the outcomes of their actions based on the state of the real environment. The novelty of our approach consists in its ability to integrate both physical and behavioural predictions into the same framework, based on heterogeneous mental models which are used to evolve internal, imaginary scenarios within the agent. We apply our generic architecture to different contexts, including artificial intelligence competitions, which require the agent using our approach to perform physical and behavioural anticipation in continuous space and time. We evaluate the applicability of our approach to realistic conditions such as noisy perception, decision time constraints and imperfect world models. Results demonstrate the genericness of our approach from one scenario to another without modifying the agent architecture, and highlight the possible uses of the proposed mental simulation framework. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Wednesday, October 28th 2015, 9:00amm, Ungar 330G Ms. Negin Arhami Department of Computer Science University of Miami will present Automated Theorem Proving by Translation to Description Logic Many Automated Theorem Proving (ATP) systems for different logical forms, and translators for translating different logical forms from one to another, have been developed and are now available. Some logical forms are more expressive than others, and it is easier to express problems in those logical forms. However, their ATP systems are relatively new, and need more development and testing in order to solve more problems in a reasonable time. To benefit from the available tools to solve more problems, different ATP systems and translators can be combined. An experiment has been designed and carried out to compare all different possible ways of trying to solve a problem in different logical forms. Description Logic (DL) sits between Conjunctive (CNF) and Propositional Logic. Saffron, a CNF to DL translator, has been developed, which bridges the gap between CNF and DL. Moreover, Description Logic Form (DLF), a new syntax for Description Logic, has been designed. Automated Theorem Proving by translation to DL is now an alternative way of solving problems expressed in logics more expressive than DL, by combining translators from those logics to CNF, Saffron and a DL ATP system. This is a Department of Computer Science PhD Defence. Wednesday, October 21st 2015, 5:00pm, Ungar 506 Dr. Lynn Cherny School of Communication University of Miami will present Visual Topic Modeling with Networks Topic modeling is a method for analyzing latent patterns in text documents to determine what they are "about." With this method, you can relate documents to each other to find similar themes. I'll walk you through a demo that uses surprisingly little code (just a bit of Python) to analyze a document set and produce a visual interactive network diagram of related topics using the open source network analysis tool Gephi. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Wednesday, October 14th 2015, 5:00pm, Ungar 506 Juan Artigas University of Miami Information Technology University of Miami will present UMIT Clinical Applications Systems A Brief Overview and and Opportunities The University of Miami's Information Technology Clinical Applications Systems (CAS) team is charged with the following mission: To strategically implement and utilize information technology in partnership with innovation in clinical practice, research and discovery, and education to positively and substantively transform the future of healthcare and to improve patient care, outcomes, and satisfaction. A brief overview of all our systems will be given, as well as some opportunities for internships and paid positions. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Wednesday, October 7th 2015, 5:00pm, Ungar 506 Dr. Elizabeth Losin Department of Psychology University of Miami will present Towards a Culturally Informed Social Neuroscience: How Computer Science can Help Untangle Relationships Between Culture, Brain and Health To make progress toward an integrative understanding of human brain - and the role it plays in governing our diverse behaviors and intricate social organization - it is crucial to consider cognition, behavior, and their neural precursors in the context of cultural experience. It is also critical to have the proper computational tools to tackle these complex questions. I will describe a series of studies focused on three main questions: How is culture acquired (sociocultural learning)? How does culture shape the brain once acquired (sociocultural plasticity)? And how do these processes influence psychological and neural mechanisms related to health? I will highlight how the computational tools of computer science can help us tackle these complex questions with increased efficacy and rigor. Specific studies address the neural underpinnings of imitative learning biases related to gender, race and political ideology, and sociocultural influences on the neural and psychological mechanisms of pain perception. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Wednesday, September 30th 2015, 5:00pm, Ungar 506 Joseph Masterjohn will present Encasement: A Method to Compute Geometric Arrangements We present an algorithm for the robust and efficient computation of arrangements of curves in 2D. Computing arrangements of curves in two dimensions and in general curves and surfaces in three dimensions is a fundamental procedure that stands as the basis for numerous other algorithms in computational geometry. An arrangement of a set of curves in two dimensions is a decomposition of the plane into connected cells, that is induced by the curves. The arrangement is described by its features: vertices (intersections of curves), edges (segments of curves), and faces (space that is bounded by segments of curves). An encasement is a set of convex polygons that isolate the features of the arrangement. We leverage techniques that have been proven effective in manipulating linear geometry and solving univariate polynomials in order to construct arrangements of non-linear geometry, specifically implicit algebraic curves whose parameters are provided by multivariate polynomials. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Wednesday, September 16th 2015, 5:00pm, Ungar 506 Google will present Computer Science Applied: A Brief Google Sampler Crawling the web, navigating the globe, personalized recommendations, speech recognition - Google's exciting projects go far beyond the core search product it started as. But how do these areas connect to the topics in your computer science and engineering courses? With examples from products like Google Maps and Search, join us as we discuss some of the computer science concepts that apply to organizing the world's information and making it universally accessible and useful. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Wednesday, September 16th 2015, 5:00pm, Ungar 506 Luis Cardona and Colleagues Nielsen will present How Nielsen Uses Technology and Computer Science Tools to Help Clients Win Nielsen is the world's largest market research company. In order to maintain its competitive edge, Nielsen invests heavily in its technology infrastructure, global acquisitions, and its people. One of the most notable investments is in the Emerging Leaders Program (ELP) and Emerging Technologists Program (ETP). Here, they take individuals with math, computer science, operations, engineering, and technology backgrounds and expose them to different areas of Nielsen. We will present example projects we have worked on at Nielsen, and how these are facilitated by Technology and Computer Science tools. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Wednesday, September 9th 2015, 5:00pm, Ungar 506 Dr. Hassan Al-Ali The Miami Project to Cure Paralysis Miller School of Medicine University of Miami will present Drug Target Deconvolution using Machine Learning and Information Theory We combined experimental approaches from Biology with machine learning approaches from Computer Science to make progress in Drug discovery. Currently, drug discovery is dominated by two main techniques: 1-biochemical screening and 2- cellular screening. Biochemical screening tests millions of compounds with a single biological target (ex. enzyme), and is therefore very efficient at identifying inhibitors for that target. However, it does not inform us on off-target effects of these inhibitors. Cellular screening, on the other hand, tests compounds on whole cells, and can therefore identify compounds with desirable overall effects. Unfortunately, it is difficult to understand the mechanisms behind these effects, since the exact molecular targets remain unknown. To address this problem, we combined the two drug discovery approaches through the use of learning algorithms. We used support vector machines and information theory to relate the biochemical effects of compounds (on hundreds of individual targets) to their cellular effects. This enabled us to identify possible drug targets for enhancing axon regeneration. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Wednesday, September 2nd 2015, 5:00pm, Ungar 506 Dr. Kevin Collins Department of Computer Science University of Miami will present A Comprehensive Analysis of the Worm Circuitry that drives Egg-Laying Behavior We describe the circuitry that drives worm egg laying behavior, using optical approaches to record and manipulate activity in behaving animals. Because egg laying shows features of a two-state Markov chain, our studies may reveal how such behavior is generated by biological circuits. Our data show the information flow between cells, illustrating how external and internal sensory information regulates cell signaling to promote the onset and termination of the behavior. The data should be amenable to tools from computer science that model the circuitry and make predictions about how non-linear sensory information is integratedto drive a simple binary behavior state. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Wednesday, August 26th 2015, 5:00pm, Ungar 506 Geoff Sutcliffe Department of Computer Science University of Miami will present Things You Can't Do with a Vampire The Vampire Automated Theorem Proving (ATP) system has been very successful at proving theorems in first-order logic. Vampire has won the First-Order Form (FOF) division of the CADE ATP System Competition (CASC) for all except one of the last 15 years, and many papers highlighting Vampire's strengths have been published. This talk examines the flip side of the Vampire coin ... what kinds of problems are difficult or even impossible for the latest incarnation of Vampire. The talk will help users decide when to use Vampire, and when to use another ATP system, will help the Vampire developers direct their work, and provides the data required to build a portfolio ATP system with Vampire as a component. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served. Monday 27th April 2015, 5:00pm, UB230 Dr. Wenwen Dou Department of Computer Science University of North Carolina will present Interactive Visual Analysis of Large Text Corpora Through Topic Modeling and Event Identification The increasing amount of textual data bears valuable insights in domains including business intelligence, science impact, and digital humanities. While automated text-analysis algorithms produce compelling results on summarizing and mining textual data, the end results are often too complex for average users to make decisions upon. In this talk, I will introduce my research efforts on integrating automated data-analysis algorithms with visual analytics systems that help decision makers make sense of large-scale textual data interactively. My research promotes topical trends and event structures as essential ingredients in depicting large text collections. Events, which are characterized by changes in topics, location, people, and time, are key elements for decision makers to understand, analyze, and even respond to certain activities (e.g. social movements, marketing campaigns). I will introduce systems that integrate text-analysis algorithms with human assessment of the extracted topics and events. To scale up my approach, I will present an architecture that takes advantage of parallel processing and cloud computing methods to enable the collection and analysis of vast amount of textual data. To further incorporate existing knowledge into topic models, I will introduce new approaches that leverage keywords/tags to shape the formation of topics and events. Through integrating statistical methods, the scalable architecture, and interactive visualizations, we provide an analysis environment for uncovering hidden topics and events, and extracting actionable insights from text corpora. Presented by the Department of Computer Science. Wednesday 22nd April 2015, 5:00pm, UB506 Luis Olazabal Department of Computer Science University of Miami will present UHealth iOS Mobile Application The University of Miami hired a third party company to build a University of Miami UHealth Mobile application for iOS and Android. On September 2014, I joined the UHealth IT department as an intern. My original task was updating and adding new features to the iOS Application. Besides programming I was allowed to attend weekly Project Management meetings in which the different teams of the UHealth IT department provided their weekly feedback. During these meetings project managers also went over the process of launching their application. As an intern, I was able to perform my task as iOS developer, but also got to experience the management and business side of Information Technology. Alfonso Curbelo Jimenez Department of Computer Science University of Miami will present Working at NextU This report describes the internship I spent at Next University Inc. and presents several tasks completed during this time, such as: - Implementing a plugin named Code Editor for Moodle (The learning management system used at NextU). - Creating a profile page for the students. - Creating a MongoDB and populating it with Moodle's database. - Generating a JSON file with the MongoDB's data. - Creating a web service that could create, update and enroll users into courses. - Replacing Moodle's default SMTP service with Mandrill (a third party tool to send emails). - Creating a plugin that generates a certificate when a student completes a program. - Generating several tables and graphs with SQL and Periscope.io to analyze students' data. - Creating a plugin that stores how many active/suspended users NextU has per month. - Fixing several issues in the process. In order to complete these tasks I had to use several programming languages such as Java, PHP, JavaScript and SQL, as well as different markup languages like HTML, XML and CSS. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 8th April 2015, 5:00pm, UB506 Dr. Victor Milenkovic Department of Computer Science University of Miami will present Automated Characterization of Critical Angles for a 4DOF Robot Consider a polyhedral obstacle and a polyhedral robot that translates in three dimensions and rotates about an axis. For example, a drone navigating a warehouse. Construction of the configuration space would permit rapid planning of flight paths for the drone. The 4D configuration space of the robot can be considered to be a 3D configuration space, which is a Minkowski sum of polyhedra, that varies with angle. The 3D space changes combinatorial structure at critical angles. Mayer, Fogel, and Halperin compute the critical angles and the structure changes for convex obstacle and robot (Solid Modeling 2010, CAD 2011). We extend this work to general polyhedra. Critical angle is characterized set of contacts between robot and obstacle vertices, edges, and facets. The angles where a case occurs are the roots of a univariate polynomial of degree at most 6. There are 30,362 distinct special cases that result when some of the contacting pairs share vertices. Factoring the polynomials that arise during construction would be prohibitively expensive, yet knowing the factors is a requirement for any numerical approach. This talk gives an offline algorithm for factoring the polynomials and building lookup tables. Using a few lookups and a fast online algorithm, the construction can determine the factors of any criticality polynomial and can identify ones that have already been solved. Joint work with Elisha Sacks and Nabeel Butt of Purdue University This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 1st April 2015, 5:00pm, UB506 Dr. Chaoming Song University of Miami will present Quantifying Social Dynamics: From Scientific Impact to Social Escalation The emergent processes driving social dynamics are a product of complex interactions among large numbers of individuals. In this talk, I will show several examples of complex human dynamics across various social systems, from scientific impact to social escalations. In the first part, I demonstrate a mechanistic model for the citation dynamics of individual papers, allowing us to collapse the citation histories of papers from different journals and disciplines into a single curve, indicating that all papers tend to follow the same universal temporal pattern. Next we focus on sudden uprisings occurred in social unrests, providing evidence instead of a remarkable gestational phase marked by self-organized aggregation through Social Networking Sites. Development hyper-escalates ahead of an uprising, enabling prediction of the real-world onset with substantial lead-times. Following a close parallel with the correlation clustering in quasi-particles and cyber-enabled contentious politics, we develop a theory of multi-agent adaptation agnostic of country and language. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 25th March 2015, 5:00pm, UB506 Mohamed Sordo will present Knowledge Extraction from text for Music Recommendation A large portion of the knowledge contained in the web is stored in unstructured natural language text. In order to acquire and formalize this heterogeneous knowledge, methods that automatically process this information are in demand. Knowledge Extraction can be defined as the process of acquiring knowledge from unstructured or structured data. This knowledge is represented in a machine-readable format, which can enable a computer system to perform complex tasks such as inferring new facts from the data or communicate with an end user in natural language. This talk will focus on a specific application of Knowledge Extraction from the web, namely for Music Recommendation. First, a brief overview of Knowledge Extraction systems is given. Then, a method to acquire knowledge from musical texts is presented. The extracted knowledge is represented as a graph, from which the recommendations are computed. A major advantage of this approach is that the recommendations can be conveyed to the user using natural language, thus providing an enhanced user experience. The rest of the talk will focus on measures to assess the validity of the extracted knowledge graph from a linguistic and a knowledge perspective, and will briefly explore the richness of the explanations given to the user as to why a musical item has been recommended. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Monday 23rd March 2015, 5:00pm, UB506 Nesreen Ahmed Purdue University will present Large-scale Network Analysis and Graph Mining Network analysis and graph mining play a prominent role in providing insights and studying phenomena across various domains, including social, behavioral, biological, transportation, entertainment, and financial domains. For example, online activity and interaction networks formed from electronic communication (e.g., email), social media (e.g., Twitter), and content sharing (e.g., Facebook). A key characteristic of these networks is that their complex structure is massive and continuously evolving over time, which makes it challenging and computationally intensive to analyze, query, and model these networks in their entirety. In this talk, I will discuss challenges of large-scale network analysis and show how sampling can mediate between the accuracy requirements of the network analysis task, the network characteristics, and the available resources. The talk will include a discussion of sampling, estimation, and visual network analytics and mining, showing how to use them for high-impact machine learning and big data applications in social, biological, and behavioral domains. Presented by the Department of Computer Science. Thursday 19th March 2015, 5:00pm, UB230 Dr. Luis Ortiz Stony Brook University will present On Networks and Behavior: Strategic Inference and Machine Learning Studying complex behavior in economic, social, or other similar systems is an important scientific endeavor with potentially direct impact to society via the eventual commercialization of relevant technology. The big-data revolution offers the opportunity to easily collect and process large amounts of data recording system behavior. Yet, our fundamental understanding of real-world complex systems remains slim at best. In this talk, I will summarize research from my group that takes a modern AI, machine learning, and engineering approach to questions about systems in domains where global behavior results from complex local interactions of agents embedded in a network. Our particular interest is interactions resulting from the distributed reasoning and the deliberate decisions of a large number of agents (e.g., a social network). In our work, we seek, and provide, algorithms that scale polynomially with the number of agents and thus can deal with relatively large systems. I will also illustrate our approach to causal strategic inference and to machine learning from strictly behavioral data in a real-world domain: the U.S. Congress. I will conclude with a summary of my broader research plan and my overall contributions, including applications to risk analysis in interdependent-security systems and, time permitting, spatio-temporal prediction from sparse measurements of cellular-data loads at base stations/cell towers, wireless localization, and graphical models for competitive networked economies Presented by the Department of Computer Science. Wednesday 18th March 2015, 5:00pm, UB506 Google will present Engineering at Google Scale Google executes over 3.5 billion searches daily, each involving 1000 computers over dozens of locations around the world. The average duration is 250 milliseconds - a double blink of the eye. It isn't just about Web pages, 300 hours of video are uploaded to YouTube every minute. Android powers hundreds of millions of mobile devices in more than 190 countries around the world, over 25 million sites are connected to Google Analytics, and the list goes on. Join us to dig into what it means to organize the world's information to make it universally accessible and useful - and the enormous engineering challenges that brings. Whether it's searches per day or miles driven using Google Navigation, learn about how we operate at a scale the world has never seen. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Tuesday 17th March 2015, 5:00pm, UB506 Dr. Aviv Segev Korea Advanced Institute of Science and Technology Visiting Scientist at MIT will present From Knowledge Modeling to Service Creation and Research Prediction The world today is characterized by too much knowledge. With so much knowledge readily accessible, the question is: what can we do with it. I will describe methods to extract and model knowledge so that it can be deployed effectively as services. I will show how knowledge extraction can be used for trend analysis and for predictions from technology and from academic publications based on social network methods. In addition, I will present implementations of methods of large scale knowledge manipulation in fields such as information for video understanding, knowledge maps for learning, and emergency crisis response. Presented by the Department of Computer Science. Monday 16th March 2015, 5:00pm, UB506 Dr. Surajit Chaudhuri Microsoft Research will present Information at Your Fingertips: Only a Dream for Enterprises? In the last decade, the information worker at enterprises has benefited from the evolution of technology in scalable data platforms, ad-hoc data analysis as well as data visualization. While such accomplishments have advanced the state of the art of enterprise analytics, relatively little progress has been made in the critical area of information finding. We will discuss why this has been the case and outline potential opportunities and exploratory directions. Surajit Chaudhuri is a Distinguished Scientist at Microsoft Research and leads the Data Management, Exploration and Mining group. In addition as a Deputy Managing Director of MSR Redmond Lab, he also has oversight of Distributed Systems, Networking, Security, Programming languages and Software Engineering groups. He serves on the Senior Leadership Team of the Cloud and Enterprises Division. H is current areas of interest are enterprise data analytics, data discovery, self-manageability and cloud database services. Working with his colleagues in Microsoft Research, he helped incorporate the Index Tuning Wizard (and subsequently Database Engine Tuning Advisor) and data cleaning technology into Microsoft SQL Server. Surajit is an ACM Fellow, a recipient of the ACM SIGMOD Edgar F. Codd Innovations Award, ACM SIGMOD Contributions Award, a VLDB 10 year Best Paper Award, and an IEEE Data Engineering Influential Paper Award. Surajit received his Ph.D. from Stanford University in 1992. Presented by the Graduate School and the Department of Computer Science. Wednesday 4th March 2015, 5:00pm, UB506 Dr. Aritra Dasgupta NYU Polytechnic School of Engineering will present Information Visualization for Human Centered Data Exploration Information visualization techniques leverage the high bandwidth of our perceptual system for communicating key insights from data. Often, these insights stem from analysis questions which are not known apriori, but evolve in the course of analysts' interactions with the visualization. However, facilitating an implicit dialogue between the analyst (e.g. "show me the interesting patterns") and the visualization (e.g. "this combination of dimensions might be of interest to you") is inherently challenging, especially in the face of the ever-growing complexity and heterogeneity of real-world data. The overarching goal of my research has been to bridge the gap between an analyst's mental model and the visual representation of data, by studying how systematic interaction mechanisms and perceptually motivated models can benefit domain experts during data exploration. For this purpose, I have also conducted quantitative and qualitative studies with domain experts for understanding how visual design choices impact real-world data analysis and visual communication goals. Following a top-down approach, I have aimed to translate the high-level abstract data analysis questions into concrete visual tasks (e.g., detect anomaly, compare, correlate, etc.) that help domain experts gain insight into complex data. For example, the systems my colleagues and I have developed, help climate scientists gain insights from multifaceted, high-dimensional data exploration, by seamlessly switching between different levels-of-detail and analytical perspectives. While comparing climate model parameters and outputs, the systems help them understand: /when, where, and why/, models are similar or different, at different granularity of space and time, and with respect to the multidimensional parameter spaces. Adopting a complementary bottom-up approach, I have developed conceptual models that help control the effects of visual mapping parameters (e.g., position, size, color, etc.) and screen-space features (e.g., saliency, clutter, etc.) for building optimal visual representations of complex data. The models help develop metrics for quantifying information loss, visual saliency, etc., and help guide analysts towards detecting interesting patterns and anomalies. For example, while analyzing large, high-dimensional data, visual feedback from the system using the metrics, help analysts interact with various combinations of subsets of records and dimensions, where hidden patterns can be detected. By gaining this dual perspective on building visualizations, I have been able to better reflect on how to augment visualization tools with features that better exploit our visual thinking capabilities, and thereby find solutions to real-world analysis problems. I will present illustrations, demos and case studies with domain experts, such as climatologists and environmental scientists, for supporting my discussion. Presented by the Department of Computer Science. Monday 2nd March 2015, 5:00pm, UB506 Dr. Paul Parsons Western University will present Design of Interactive Visualization Tools for Complex Activities Interactive visualization tools are being used with increasing frequency to support complex activities in a number of domains, including science, healthcare, medicine, education, and others. Visualizations are representations of data and information that sit at the visual interface of computational tools, supporting the perceptual and cognitive tasks of users as they work with the underlying data to achieve their goals. Previous research has clearly shown that visualizations can significantly enhance the performance of all kinds of data-driven activities. Furthermore, it is well known that making visualizations interactive can increase their expressiveness and overall utility. Interactive visualizations allow for a human-information discourse that---if designed well---supports the performance of complex tasks and activities. One problem in this relatively young area is that much design is done in an ad hoc fashion---that is, visualization and interaction techniques are often designed for particular users, tasks, and contexts of use. This makes it very difficult to engage in systematic design practices, and to have a scientific foundation for research in this interdisciplinary area. In this talk I will present a number of related frameworks I have developed in recent years to address this problem, and to enable and support systematic, scientific design practices. I will discuss how this research provides a general, comprehensive foundation on which visualization tools in many different domains can be systematically designed. Presented by the Department of Computer Science. Friday 27th February 2015, 1:30pm, UB506 Dr. Huy Vo Center for Urban Science and Progress New York University will present Visual Exploration of Big Urban Data About half of humanity lives in urban environments today and that number will grow to 80% by the middle of this century. Cities are thus the loci of resource consumption, of economic activity, and of innovation; they are the cause of our looming sustainability problems but also where those problems must be solved. Data, along with visualization and analytics can help significantly in finding these solutions. This talk will discuss the challenges of visual exploration of big urban data; and showcase approaches taken in a study of New York City taxi trips. Taxis are valuable sensors and can provide unprecedented insight into many different aspects of city life. But analyzing these data presents many challenges. The data are complex, containing geographical and temporal components in addition to multiple variables associated with each trip. Consequently, it is hard to specify exploratory queries and to perform comparative analyses. This problem is largely due to the size of the data. There are almost a billion records of taxi trips collected in a 5-year period. TaxiVis, a tool that allows domain experts to visually query taxi trips at an interactive speed and performing tasks that were unattainable before, will be presented. Also key contributions in this work: the visual querying model and novel indexing scheme for spatio-temporal datasets will be discussed. Presented by the Department of Computer Science. Wednesday 25th February 2015, 5:00pm, UB506 Christine Lisetti School of Computing and Information Sciences Florida International University will present What Can Your Personal Virtual Health Agent Do for You? Recent epidemics of behavioral issues are calling for urgent preventive measures. Lack of trained healthcare and medical personnel, however, is leading researchers to call for computing technologies to automate some of the supportive interventions that people need, and increase their access to under-served populations. We discuss our approach to develop a novel modality for the computer-delivery of behavior change interventions via personal virtual health agents (VHA). VHAs are multimodal intelligent virtual agents who empathically deliver health interventions while adapting, in real-time, their verbal and non-verbal communication messages to those of their user's to increase engagement. We discuss the evaluation of our virtual health agents, and conclude with a discussion on design decisions to build "on-demand" personal virtual health agents as health and well-being helpers. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 18th February 2015, 5:00pm, UB506 William Jacobs Science and Engineering Librarian University of Miami will present More Effective Research using Library Resources As a Computer Scientist, you will need to go beyond textbooks to the wider world of the scientific and technical literature. Much of this material is visible on the open web, but using Google it is easy to waste time with inefficient searches and miss important information. This talk will introduce you to the wealth of Computer Science reference material, scholarly journals, books and more made available by the University of Miami Libraries. Bill Jacobs, the science librarian, will cover the most effective ways to search the library's collection and how to access the full text from on and off campus. He will also show you tools to organize and share the information you find in your literature searches and easily create bibliographies. In addition, he will briefly discuss open access and authors' rights issues. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 11th February 2015, 5:00pm, UB506 Zoey Chen School of Business University of Miami will present Social Considerations in Online Communication Word of mouth (WOM) - product information that's passed on from one person to another - has long been regarded as one of the most valuable sources of information for consumers (Brown and Reingen 1987). Unlike traditional offline word of mouth, which typically occurs among people who know each other, online word of mouth occurs among strangers who do not know, and are unlikely to ever know, one another. While it is reasonable to assume that social concerns, such as establishing or maintaining relationships, are likely to influence people's behavior offline, it is unclear whether social concerns dictate people's online word of mouth behavior. My research draws on data from large-scale field studies (e.g., analysis of over 65,000 Yelp reviews) as well as laboratory experiments to show that social considerations permeate every step of the word of mouth process. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 4th February 2015, 5:00pm, UB506 Dr Clayton Ewing School of Communication University of Miami will present Games: Moving Beyond Entertainment Games are largely seen as a source of entertainment but their ability to engage people on deeper levels make them an ideal medium for activism, advocacy, education and research. Using examples from my own work and other leading practitioners in the field, I will give an overview of games that are leading the way to change the way we think about games in the world. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 28th January 2015, 5:00pm, UB506 Dr Juhong Park School of Architecture University of Miami will present Design | Machine | Learning In The Architecture Machine, Negroponte (1973) presented a vision of an evolutionary design process in which designers propose an idea to a machine and the machine also trains designers. In this mutual manner, he proposed an interrelated design process between two distinct species (human and computer) and two intelligences (the designer and the architecture machine). This seminar introduces architectural research on machine learning in design and design education that slowly but surely illustrates how his dream becomes reality. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 21st January 2015, 5:00pm, UB506 Dr Angie Laird Departments of Physics, Psychology, and Neuroscience Florida International University will present What Can Computational Approaches and Databases Tell Us About "Big Data" in Cognitive Neuroimaging? Human functional magnetic resonance imaging (fMRI) is an experimental discipline that establishes structure-function correspondences in the brain through the interdisciplinary application of principles from physics/engineering, computer science, statistics, biology, medicine, and psychology. In the last two decades, research in fMRI has resulted in an enormous amount of data intended to spatially localize the neural regions that perform specific mental operations (e.g., perception, action, cognition, emotion and autonomic functions). I will describe the fMR imaging modality and present prior work developing novel meta-analysis methods for investigating the spatial topography of distinct brain regions, as well as various computational techniques for mining a large database of neuroimaging data to more fully understand large-scale intrinsic connectivity networks and the functional organization of the human brain. Lastly, I will present recent results demonstrating agreement between data mining results and existing cognitive models of human face perception and the cognitive control of emotional processing. Taken together, our results suggest differential recruitment of task-dependent networks to enable cognitive, perceptual, and affective processing in the context of distinct goal-directed behaviors. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 3rd December 2014, 5:00pm, UB230 Project Students Department of Computer Science University of Miami will present Project Reports Pedro A. Pena: Developing Educational Apps at Tinybop, Inc. Marcos Corea: Documentation and Tutorial for the Unity Game Engine Mickey Fried: Learning from and Applying the Unity Game Engine Tutorial This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 19th November 2014, 5:00pm, UB230 Stephan Schürer Department of Pharmacology University of Miami will present Scalable Informatics and Knowledge Management across Disciplines to enable Discovery in the Era of "Big Data" Huge datasets in areas of chemical biology, omics, systems biology, chemistry and other areas are generated and published via many large-scale publicly funded projects. But these data are often difficult to analyze, integrate, and re-use for discovery projects for which they were not originally intended. Here I illustrate some of our efforts to address these challenges. We developed BioAssay Ontology (BAO) to enable the description and modeling of chemical biology screening data in PubChem. BAO is now used on several organizations and projects including in the BioAssay Research Database (BARD), a public chemical biology and cheminformatic resource. In the Library of Integrated Network-based Cellular Signatures (LINCS) program, we are developing an integrated knowledge environment to enable data modeling across large and diverse systems-level drug response signatures from many sources. In the LINCS pilot phase we have built the LIFE search engine that uses a semantic knowledge model and other innovations for fast retrieval of results related to query terms and concepts. Formal ontologies can also be applied to describe and model chemical functional groups and chemical reactivity. Combining systems-level disease models, drug-target interaction predictions, and efficient computational exploration of synthetically accessible chemistry space will enable us to identify opportunity spaces for novel poly-pharmacology drugs. We are applying this approach to prioritize novel dual kinase and bromodomain inhibitor scaffolds for various cancers and other diseases. In one example we have recently identified novel dual EGFR BRD4 compounds entirely based on computational models. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 12th November 2014, 5:00pm, UB230 Rajeev Prabhakar Department of Chemistry University of Miami will present Computer Simulations of Chemical/Biochemical Reactions The selective hydrolysis of the extremely stable peptide or amide bond of peptides and proteins is required in a wide range of biological, biotechnological and industrial applications such as control of the cell cycle, cell death, proteomics, protein engineering, designing of catalytic drugs and bioethanol production. We have applied innovative theoretical techniques including molecular dynamic (MD) simulations, quantum mechanics (QM) and quantum mechanics/molecular mechanics (QM/MM) to investigate the mechanisms of aspartyl protease (beta-secretase) and mono- and binuclear metal center containing enzymes (insulin degrading enzyme and leucine aminopeptidase respectively) and their synthetic analogues. In the first step, we have investigated the roles of metal ions, ligands, second coordination shell residues and the protein environment in the catalytic functioning of enzymes. In the next step, the knowledge of the roles of metal ions and the ligand environment in the mechanisms of existing synthetic analogues (Pd(II), Cu(II) and Co(III) metal complexes) of enzymes will be combined with the information acquired in the first step to design more efficient synthetic analogues of peptide cleaving enzymes. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 5th November 2014, 5:00pm, UB230 Dr. Sandro Andrade Department of Finance University of Miami will present Analyst Coverage, Information, and Bubbles We examine the 2007 stock market bubble in China. Using multiple measures of bubble intensity for each stock, we find significantly smaller bubbles in stocks for which there is greater analyst coverage. We further show that the abating effect of analyst coverage on bubble intensity is weaker when there is greater disagreement among analysts. This suggests that, in line with resale option theories of bubbles, one channel through which analyst coverage may mitigate bubbles is by coordinating investors' beliefs and thus reducing its dispersion. Stock turnover provides further evidence consistent with this particular information mechanism. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 29th October 2014, 5:00pm, UB230 Dr. Bonnie Kirkpatrick Department of Computer Science University of Miami will present Science as a Conversation We will follow the development of hidden Markov models and graphical models in machine learning and discuss how they relate to the pedigree likelihood model in family genetics. This is a scientific conversation that has been going on for over a quarter of a century between statisticians, computer scientists, and geneticists. The conclusion to this conversation has yet to be discovered. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 22nd October 2014, 5:00pm, UB230 Dr. Enrico Bertini Department of Computer Science and Engineering NYU Polytechnic School of Engineering will present Visualization as Intelligence Amplification Visualization has been around now for a good amount of time. IEEE Vis, the premier academic conference in this area, has been founded in 1990. The foundational book "Readings in Information Visualization" has been published in 1999. Tableau Software has been founded in 2013. I think it's time to stop for a moment and think. What is the role of visualization? What do we want to build in the future? What are the main challenges and obstacles we will be facing? I take this opportunity to reflect about these questions and the role of visualization in research and in the society at large. Drawing from Douglas Engelbart's "Augmenting Human Intellect", I argue that the main goal of Visualization is to "amplify our intelligence" and show examples of systems that achieve this goal. During the talk I will showcase research we are conducting in my lab at New York University and will discuss challenges and ways we can make visualization more useful and pervasive. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 8th October 2014, 5:00pm, UB230 Iman Saleh Moustafa Department of Computer Science University of Miami will present Formal Specification and Verification of Data-Centric Web Services Web Services are reusable software components that make use of standardized interfaces to enable loosely-coupled business-to-business and customer-to-business interactions over the Web. In such environments, service consumers depend heavily on the service interface specification to discover, invoke, and synthesize services over the Web. Data-centric Web services are services whose behavior is determined by their interactions with a repository of stored data. A major challenge in this domain is interpreting the data that must be marshaled between consumer and producer systems. While the Web Services Description Language (WSDL) is currently the de facto standard for Web services, it only specifies a service operation in terms of its syntactical inputs and outputs; it does not provide a means for specifying the underlying data model, nor does it specify how a service invocation affects the data. The lack of data specification potentially leads to erroneous use of the service by a consumer. In this work, we propose a formal contract for data-centric Web services. The goal is to formally and unambiguously specify the service behavior in terms of its underlying data model and data interactions. We address the specification of a single service, a flow of services interacting with a single data store, and also the specification of distributed transactions involving multiple Web services interacting with different autonomous data stores. We use the proposed formal contract to decrease ambiguity about a service behavior, to fully verify a composition of services, and to guarantee correctness and data integrity properties within a transactional composition of services. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 24th September 2014, 5:00pm, UB230 Henry Senyondo Department of Electrical and Computer Engineering University of Miami will present Modeling and Verification of Cyber-Physical Systems for Smart Grid Infrastructure A shift in the power system infrastructure from a static to a more dynamically distributed environment has contributed towards the need for the enhancements of the future cyber physical support systems. The power systems have evolved from a unidirectional to a bidirectional infrastructure with millions of nodes from the source to the destined power user. The users could be storage facilities or applications. The monitoring of this dynamic network involves ensuring that the network is in a stable state under all circumstances that could include natural disasters, attacks from terrorist activities, undetected malfunctions and poor configurations. The existing security schemes in power control systems only consider securing the power grid at single point of the infrastructure level. The current goal of this work is to develop tools that could support this infrastructure and could ensure a reinstatement in case of such disasters. This would require both securing the infrastructure at the hardware level and at the application level. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wednesday 17th September 2014, 5:00pm, UB230 Kyle Poore Department of Computer Science University of Miami will present Single- and Multi-Channel Whistle Recognition with NAO Robots We propose two real-time sound recognition approaches that are able to distinguish a predefined whistle sound on a NAO robot in various noisy environments. The approaches use one, two, and four microphone channels of a NAO robot. The first approach is based on a frequency/band-pass filter whereas the second approach is based on logistic regression. We conducted experiments in six different settings varying the noise level of both the surrounding environment and the robot itself. The results show that the robot will be able to identify the whistle reliably even in very noisy environments. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wedensday 10th September 2014, 5:00pm, UB230 Dr. Charles A. Kamhoua Air Force Research Laboratory will present Security and Interdependency in a Public Interactions Cloud: A Game Theoretic Approach As cloud computing thrives, many organizations, both large and small, are joining a public cloud to take advantage of its multiple benefits. Especially public cloud based computing, is cost efficient, i.e., a cloud user can reduce spending on technology infrastructure and have easy access to their information without up-front or long-term commitment of resources. Despite those benefits, concern over cyber security is the main reason many large organizations with sensitive information such as the Department of Defense have been reluctant to join a public cloud. This is because different public cloud users share a common platform such as the hypervisor. An attacker can compromise a virtual machine (VM) to launch an attack on the hypervisor which, if compromised, can instantly yield the compromising of all the VMs running on top of that hypervisor. This work shows that there are multiple Nash equilibria of the public cloud security game. However, the players use a Nash equilibrium profile depending on the probability that the hypervisor is compromised given a successful attack on a user and the total expense required to invest in security. Finally, there is no Nash equilibrium in which all the users in a public cloud fully invest in security. This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Wedensday 3rd September 2014, 5:00pm, UB230 Prof. Neil Johnson Department of Physics University of Miami will present Shining Computational Light on the Complex Nature Of Human Interactions This talk shows how computational analysis being performed within UM's new cross-disciplinary Complexity cluster is shedding new light on collective human behavior, challenging existing beliefs and revealing a number of surprises. Computational analysis through data-mining, simulation, and traditional computation, together with collaborations across departments, are playing a key role in unraveling both the complexity of the data and its meaning. As a result, the possibility emerges to predict certain types of human behavior, through a combination of machine-learning of physical cues with simple mathematical rules that can benchmark human interactions. Potential applications of the analysis are discussed, which range from nationalsecurity through to psychology and public health, and from finance through to many-body soft matter behaviors in physics and chemistry. Ongoing collaboration with UM faculty members Prof. Stefan Wuchty (Computer Science and Complexity cluster), Prof. Chaoming Song (Physics and Complexity cluster), Prof. Daniel Messinger (Psychology), Prof. Elvira Maria Restrepo (Geography) and UM PhD students Pedro Manrique (Physics) Hong Qi (Physics) and Nicolas Velasquez (INS) This is another in the Department of Computer Science Pizza Seminar series. Refreshments will be served on the Ungar Building Terrace at 4:30PM Friday 25th April 2014, 3:30pm, UB411 Faisal Sikder Department of Computer Science University of Miami will present Scalable Self-Tuning Implementation of Smith-Waterman Algorithm for Shared-Memory Multicore Computers Improved version of the Smith-Waterman algorithm (SWA) is most widely used for local alignment of a pattern (or query) sequence with a Database (DB) sequence. This dynamic-programming algorithm is computation intensive. To reduce time for computing alignment score matrix, parallel versions have been implemented on GPUs and multicore CPUs. These parallel versions have shown significant speedup when compared with their corresponding sequential versions. Our initial evaluation of an OpenMP parallelization of SWA has shown linear speedup on multicore CPUs, but a closer look at performance data from both sequential and parallel versions have revealed two undesired effects: (i) As the length of the DB sequence increases, the number of elements of the alignment score matrix /H/ computed in per unit time initially increases, then reaches a maximum, and finally decreases continuously; (ii) the length of the DB sequence where decline starts is different for different CPUs. To overcome the computation rate decline we have proposed a run-time self-tuning algorithm. It determines the length, /l/, of a DB sequence that maximize computation rate during execution time. Then, divide computation of /H/ into computation of a set of submatrices, such that the number of columns in each submatrix is about /l/. Our study also found that the number of per-core-threads that delivers the highest rate of computation differs from CPU to CPU. Our proposed algorithm determines optimal number of threads during execution time and creates optimal number of threads for highest possible computation rate. Our extensive evaluations of the proposed self-tuning algorithm on three different multicore multi-CPU shared-memory machines have shown significant performance improvement. This is a Department of Computer Science Master's Thesis Defense. Thursday 3rd July 2014, 12:30pm, UB411 Justin Stoecker Department of Computer Science University of Miami will present Interactive Visualization Systems and Sensor-Based Interfaces for Robotics and Medical Data Analysis Scientific visualization enables researchers to extract understanding from data through visual representations. Interactive visualizations are particularly powerful for exploration of data, but they must build and manipulate representations in real time; this is especially difficult for large or complex data. Advances in human-computer interfaces also provide for new ways to engage with visualizations. This defense will cover visualization systems that address challenges in visual representation, real-time rendering, and natural interaction. The first part of the talk will discuss three projects in robotics: (1) dynamically generated visualizations of robot behaviors in a multi-agent environment; (2) exploring 3D robot configuration spaces; (3) interactively mapping human motions to humanoid robots with visual feedback. The second part of this talk will be on a system developed with doctors at the Oregon Health and Science University (OHSU) Hospital for surgical navigation of MR and CT images using a touchless, gesture-based interface. Our medical visualization software provides enhanced understanding of complex anatomical relationships and introduces novel interactions with volumetric data. This is a Department of Computer Science PhD Thesis Defense. Wednesday 16th April 2014, 5:00pm, UB230 (talk @ 5:00pm), Pizza in UB411 @ 4:30pm Atif Memon Department of Computer Science University of Maryland will present The First Decade of GUI Ripping: Extensions, Applications, and Broader Impacts This seminar provides a retrospective examination of GUI Ripping - reverse engineering a workflow model of the graphical user interface of a software application - born a decade ago out of recognition of the severe need for improving the then largely manual state-of-the-practice of functional GUI testing. In these last 10 years, GUI ripping has turned out to be an enabler for much research, both within our group at Maryland and other groups. Researchers have found new and unique applications of GUI ripping, ranging from measuring human performance to re-engineering legacy user interfaces. GUI ripping has also enabled large-scale experimentation involving millions of test cases, thereby helping to understand the nature of GUI faults and characteristics of test cases to detect them. It has resulted in large multi-institutional Government-sponsored research projects on test automation and benchmarking. GUI ripping tools have been ported to many platforms, including Java AWT and Swing, iOS, Android, UNO, Microsoft Windows, and web. In essence, the technology has transformed the way researchers and practitioners think about the nature of GUI testing, no longer considered a manual activity; rather, thanks largely to GUI Ripping, automation has become the primary focus of current GUI testing techniques. Atif M. Memon is an Associate Professor at the Department of Computer Science, University of Maryland, where he founded and heads the Event Driven Software Lab (EDSL). Researchers at the EDSL study issues of design, development, quality assurance, and maintenance of such software applications. He designed and developed the model-based GUI testing software GUITAR, which operates on Android, iPhone, Java Swing, .NET, Java SWT, UNO, MS Windows, and web systems, and leverages a resource cloud for test automation. He has published over 100 research articles on the topic of event driven systems, software testing, and software engineering. He is the founder of the International Workshop on TESTing Techniques & Experimentation Benchmarks for Event-Driven Software (TESTBEDS). He also helped develop the workshop on Experimental Evaluation of Software and Systems in Computer Science (EVALUATE). This is another in the Departmental Pizza Seminar series. Wednesday 16th April 2014, 3:30pm, UB411 Muhammad Nassar Department of Computer Science University of Miami will present Automated Theorem Proving using the TPTP Process Instruction Language The TPTP (Thousands of Problems for Theorem Provers) World is a well-established infrastructure for Automated Theorem Proving (ATP). In the context of the TPTP World, The TPTP Process Instruction (TPI) language makes possible the easy manipulation of logical formulae and provides better control over the use of ATP systems. A shell interpreter has been developed for the language. It is used as a standalone program to execute one TPI command at a time. This thesis presents work done building and testing the interpreter, demonstrates the use of the TPI language and the interpreter in theorem proving, discusses generic control of ATP systems, and shows how such control is achieved through an extension to the TPI language. This is a Department of Computer Science Master's Thesis Defense. Wednesday 16th April 2014, 10am, UB411 Negin Arhami Department of Computer Science University of Miami will present The Efficiency of Automated Theorem Proving by Translation to Less Expressive Logics There is a trade-off between expressivity of a logic, and the power and reliability of the available Automated Theorem Provers (ATPs) for those logics. Less expressive logics typically have powerful ATPs, but it is easier to express a problem in a more expressive logic. To solve a problem written in a given logic, an ATP for that logic can be used, or the problem can be translated to a less expressive form where a more powerful ATP can be used. If the problem is translated to a less expressive form, then again the same two possibilities are available, until no further translation is possible. In this research, an experiment has been designed and carried out to compare all these possibilities. No translator was available to translate from Conjunctive Normal Form (CNF) to Description Logic (DL). A translation procedure for translating CNF to DL, and its implementation as Saffron are also presented. This is a Department of Computer Science Master's Thesis Defense. Monday 14th April 2014, 10am, UB411 Piyali Nath Department of Computer Science University of Miami will present Distributed and Parallel Optimization of Captured Human Motions for the Generation of Robust and Stable Motions on Simulated Humanoid Robots There are several areas of research in the field of motion learning for robots. RoboCanes, a research group belonging to the Department of Computer Science at University of Miami, uses the process of recording a human motion and optimizing the same to get stable motions for a simulated robot in the framework for the agent code. The major downside of this approach is that this optimization process may take hours to return a good set of parameters which leads to the need of a fast parallelizing approach. In this thesis, we present a parallel approach that extends the existing framework to generate stable motions for simulated humanoid robots using a motion capture framework and then optimizing the motions efficiently in a distributed and parallel environment. The approach is based on a client server network of systems where the server system controls the optimization process and distributes the particles (candidate solutions of an optimization process) to multiple client systems, which are responsible for running simulations individually, using the parameters sent by the server and then evaluating the error and returning it to the server. The experiments are conducted on three different experimental setups with four motion files and three optimization algorithms. The performance of this technique is measured with varying number of clients, each of which show considerable speedup as compared to the serial process. We have conducted more than 7000 experiments and the results show that the approach is promising and we have an average speed up increase of about 0.626 with every node we add. This is a Department of Computer Science Master's Thesis Defense. Thursday 3rd April 2014, 3:15pm, UB411 Luis Núñez Department of Computer Science University of Miami will present Software Engineering in Small Projects: The Most Essential Processes Although frequently viewed as bureaucratic and inefficient practices, some software engineering processes, in particular unit testing, may prove to be not only useful but indispensable even for small projects. Many software-related startups often do not know which software engineering processes and tools are most effective or even those that are absolutely required. In addition, they usually have significant time constraints and have limited resources. As a result, it is very common for startup businesses to overlook and omit the use of many vital processes and/or tools, without realizing that such omissions could negatively impact their project, financially, at present and many years into the future. This thesis surveys and evaluates relevant business processes for software engineering in small enterprises including requirements engineering, infrastructure selection, and testing alternatives. Consequently, this work provides a decision support methodology when selecting software processes that will ultimately result in robust, reliable, scalable, and maintainable software. This work uses focused experimentation to demonstrate how to effectively leverage unit testing techniques for small projects where resources are limited. This thesis is the first step towards a repeatable process for selecting an appropriate set of processes and tools to be used for new, small-scale software projects. The results of this model are evaluated using a focused software development experiment where an existing software product (that did not initially consider formal software engineering techniques) is redeveloped to incorporate unit testing paradigms. The outcomes of this experiment include the evaluation and comparison of software quality and level of effort of the existing product as it related to the resulting product. This presentation is presented in two parts (1) a survey of the state-of-the-art criteria for selecting software engineering processes/tools and (2) the results of the experiment where recommended techniques for unit testing are re-developed within an application that was originally developed without formal software engineering techniques. This is a Department of Computer Science Master's Thesis Defense. Tuesday 1st April 2014, 3:30pm, UB411 Basar Koc Department of Computer Science University of Miami will present Lossless Compression of Images Data compression is a challenging process with important practical applications. Specialized techniques for lossy and lossless data compression have been the subject of numerous investigations during last several decades. These general data compression algorithms were used also for compression of images with considerable success. How- ever, with recent developments of new lossy and lossless data compression algorithms designed specifically for images, researchers have begun to achieve impressive gains in image compression efficiency. Lossy image compression tolerates compromising quality and information loss at the expense of compression gain. Lossless compression, in contrast, preserves image integrity fully, which is an important consideration in certain critical applications such as medical image processing. In this thesis we present a new technique for lossless compression of images based on the pseudo-distance technique proposed by Kuroki et al. We improve their algorithm by using a dynamic pseudo-distance matrix, context-models, and utilizing block-sorting transformations before the entropy coder. While their algorithm out-performed GIF by 16%, our algorithm achieves gains of 54% over GIF and 6% over PNG. We also present a parallelized implementation of our algorithm, which results in substantial gains in compression time while providing the desired compression efficiency. We demonstrate the efficiency of our compression algorithm on standard test image sets, including non-dithered and dithered color palette, and microarray images. This is a Department of Computer Science Master's Thesis Defense. Wednesday 26th March 2014, 5:00pm, UB230 (talk @ 5:00pm), Pizza in UB411 @ 4:30pm Al Monserrat Vice-President of Sales and Marketing Citrix will present Open the Doors to the World with Computer Science In today's connected world a Computer Science degree can launch you into places and roles you never expected. Many people pick their degree with little experience or knowledge about the many different roles and opportunities that exist once they complete their college studies. With technology becoming the foundation for all functions, Computer Science becomes the key that opens the door to opportunity. During his 20+ year career in Technology, Al Monserrat has leveraged his UM Computer Science degree to lead transformational work across industries ranging from high-tech to consumer products; organizations ranging from5 million to 50+ billion; and across the globe. Join as we journey through a variety of topics of your choice in a casual discussion. Al Monserrat BS '89 is responsible for the global sales, services, and partner organizations at Citrix. Collectively, his team delivers Citrix's market-leading cloud and networking solutions to support the mobile workstyle objectives of over 230,000 customers worldwide. Monserrat has carved out a reputation as a charismatic leader who works hand-in-hand with his team, partners and solution providers to develop sales strategies that have made Citrix an industry leader. This is another in the Departmental Pizza Seminar series. Tuesday 25th March 2014, 12:30pm, UB411 Yajie Hu Department of Computer Science University of Miami will present A Model-Based Music Recommendation System for Individual Users and Implicit User Groups This dissertation introduces several relevant problems on music recommendation. To make a high-qualified recommender, we evaluate feature importance for favorite song detection from two perspectives. The experiment results expose that collaborative filtering signal is the most important feature among the analyzed features. A classifier combination method is proposed to leverage several classifiers trained by different data sources to predict music genre. The complemented genre labels are used in a recommendation system for individual users on local device. The recommender takes freshness, time pattern, genre, publish year, and favor into account to make recommendations. The recommender outperforms the baseline on mostly favorite songs. We propose an adapted recommendation method to response user feedbacks and find out local optimizations to improve the recommendation quality. Furthermore, a probability-based method is proposed to make recommendations for implicit user groups by integrating individual opinions on candidate songs. This is a Department of Computer Science PhD Thesis Defense. Wednesday 19th March 2014, 5:00pm, UB230 (talk @ 5:00pm), Pizza in UB411 @ 4:30pm Greg Miller - Director of Engineering Talent Kelly Zamora- Engineering Talent Manager Joseph Cutrono - Software Engineer Ultimate Software will present Test-Driven Development The presentation will give an introduction to Test-driven development (TDD). This process hopes to accomplish higher quality code by having the developer write a failing test, then write enough code to make the test pass, and finally refactors the new code to acceptable standards. Kent Beck, who is credited with having developed or 'rediscovered' the technique, stated in 2003 that TDD encourages simple designs and inspires confidence. We will also be talking about agile development and career paths in the CS field. Ultimate Software ranked #20 in FORTUNE's Top 100 Best Companies to Work for in America in 2014. Our enterprise cloud-based software solution offers some wonderful challenges to Software Engineers and Software Test Engineers with skillsets that students at UM possess. We are located in Weston, just West of Fort Lauderdale, and are looking for students majoring in Computer Science, CIS, Software Engineering, IT and IS! To give you an idea of how truly impressive our work environment is... on the enterprise Development Team, we've given 78 offers to interns to join us full-time over the last 7 years, and 76 of those offers have been accepted. That near 100% acceptance rate is even more impressive because many of our interns have previously interned with companies such as Amazon, Microsoft, Lockheed Martin, Intel, Accenture, and IBM's Extreme Blue internship program. Once people experience working in this environment, they always stay if given the chance. Please bring a hard copy of your resume to give it to the Ultimate Software reps to be considered for an internship or post grad position with us. This is another in the Departmental Pizza Seminar series. Wednesday 5th March 2014, 5:00pm, UB230 (talk @ 5:00pm), Pizza in UB411 @ 4:30pm Dr. Teresa Przytycka National Center for Biotechnology Information National Institutes of Health will present Towards Dissecting the Impact of Variations in RNA and DNA Secondary Structures Functionality of an RNA molecule depends on its secondary structure. It is also increasingly recognized that a DNA molecule can assume alternative structures of potential functional importance. How mutations in these structures affect their function? In this talk I will approach this question from two perspectives. First, I will describe an algorithm, called remuRNA, that allows to measure changes in RNA secondary in response to single nucleotide variation on RNA sequence. Comparing the impact of common SNPs naturally occurring in populations with the impact of random point mutations, we found that structural changes introduced by commonSNPs are smaller than those introduced by random point mutations. Interestingly, mutations in the regions susceptible to formation of alternative DNA structures are common and don't appear to be selected against. We propose an explanation why this might be the case. This is another in the Departmental Pizza Seminar series. Wednesday 26th February 2014, 5:00pm, UB230 (talk @ 5:00pm), Pizza in UB411 @ 4:30pm Dr. Daria Salyakina Center for Computational Science University of Miami will present Viral Expression Associated With Gastrointestinal Adenocarcinomas In TCGA High-Throughput Sequencing Data Up to 20% of cancers worldwide are thought to be associated with microbial pathogens, including bacteria and viruses. The widely used methods of viral infection detection are usually limited to a few a priori suspected viruses in one cancer type. To our knowledge, there have not been many broad screening approaches to address this problem more comprehensively. In this study, we performed a comprehensive screening for viruses in nine common cancers using a multistep computational approach. Tumor transcriptome and genome sequencing data were available from The Cancer Genome Atlas (TCGA). Nine hundred fifty eight primary tumors in nine common cancers with poor prognosis were screened against a non-redundant database of virus sequences. DNA sequences from normal matched tissue specimens were used as controls to test whether each virus is associated with tumors. We identified human papilloma virus type 18 (HPV-18) and four human herpes viruses (HHV) types 4, 5, 6B, and 8, also known as EBV, CMV, roseola virus, and KSHV, in colon, rectal, and stomach adenocarcinomas. In total, 59% of screened gastrointestinal adenocarcinomas (GIA) were positive for at least one virus: 26% forEBV, 21% for CMV, 7% for HHV-6B, and 20% for HPV-18. Over 20% of tumors were co-infected with multiple viruses. Two viruses (EBV and CMV) were statistically significantly associated with colorectal cancers when compared to the matched healthy tissues from the same individuals (p = 0.02 and 0.03, respectively). HPV-18 was not detected in DNA, and thus, no association testing was possible. Nevertheless, HPV-18 expression patterns suggest viral integration in the host genome, consistent with the potentially oncogenic nature of HPV-18 in colorectal adenocarcinomas. The estimated counts of viral copies were below one per cell for all identified viruses and approached the detection limit. Our comprehensive screening for viruses in multiple cancer types using next-generation sequencing data clearly demonstrates the presence of viral sequences in GIA. EBV, CMV, and HPV-18 are potentially causal for GIA, although their oncogenic role is yet to be established. This is another in the Departmental Pizza Seminar series. Wednesday 19th February 2014, 5:00pm, UB230 (talk @ 5:00pm), Pizza in UB411 @ 4:30pm William Jacobs Science and Engineering Librarian University of Miami will present Using The Library As a Computer Scientist, you will need to go beyond textbooks to the wider world of the scientific and technical literature. Much of this material is visible on the open web, but using Google it is easy to waste time with inefficient searches and miss important information. This talk will introduce you to the wealth of Computer Science reference material, scholarly journals, books and more made available by the University of Miami Libraries. Bill Jacobs, the science librarian, will cover the most effective ways to search the library's collection and how to access the full text from on and off campus. He will also show you tools to organize and share the information you find in your literature searches and easily create bibliographies. This is another in the Departmental Pizza Seminar series. Thursday 13th February 2014, 5:00pm, UB230 Odelia Schwartz Dominick P. Purpura Department of Neuroscience Albert Einstein College of Medicine will present Understanding Contextual Visual Processing Using a Principled Model of Natural Image Statistics Neural processing, perception, and cognition, are dramatically influenced by the context of events in space and time. I focus on spatial context in vision as a paradigmatic example. Perceptually, spatial context can induce intriguing illusions and salience phenomena, and plays a critical role in object grouping, segmentation, and recognition. Deficits have been implicated in neurological disorders and aging. However, the principles underlying contextual effects in vision are not well understood. An appealing hypothesis suggests that neurons represent inputs in a coordinate system that is matched to the statistical structure of images in the natural environment. I describe a computational framework that is based on a well-found account of natural image statistics, and how we have used this approach to make sense of some spatial context phenomena in neuroscience and perception. I particularly highlight recent work in which we have made progress on a challenging question, namely understanding how cortical neurons process complex natural images. Studies have rarely considered natural images, focusing instead on simpler visual stimuli such as bars and gratings. In experimental collaboration, we found that contextual modulation in primary visual cortical neurons was recruited for some images but absent for others. This variability was not well explained by standard models, but could be predicted assuming that the neurons engaged in Bayesian inference about the structure of visual inputs. Surround modulation was flexible, only being recruited when the image was inferred to contain dependencies, and muted otherwise. Our results reveal a gating of a basic, widespread cortical computation, by optimal inference about the statistics of natural input. I also discuss the broader applicability of this framework to hierarchical neural processing, deep learning of image structure in computer science, and other forms of contextual influences. Presented by the Department of Computer Science. Friday 7th February 2014, 5:00pm, UB230 (talk @ 5:00pm), Pizza in UB411 @ 4:30pm Andrew Lovett Universtät Bremen will present Modeling Visual Problem-Solving and Spatial Reasoning Visual problem-solving tasks offer important insights into how people represent and reason about space. For example, in the Visual Oddity Task participants select the image that lacks a key spatial feature. It is critical that we understand these tasks, so we can train students to be better spatial thinkers and improve achievement in science and engineering. I use computational models as tools for studying these tasks. Each model is an autonomous agent which constructs representations from 2D drawings and reasons over them to solve a problem. The models are developed based on the current understanding of human spatial reasoning. By manipulating the models' parameters and comparing their performance to humans, one can evaluate psychological theories and derive new hypotheses. In addition, the models can be integrated into future intelligent tutoring systems to improve spatial ability. Presented by the Department of Computer Science. Wednesday 5th February 2014, 5:00pm, UB230 (talk @ 5:00pm), Pizza in UB411 @ 4:30pm Dr. Jong-Hoon Kim School of Computing and Information Sciences Florida International University will present A Telepresence Humanoid Police Robot for Disabled Veteran and Police Officer To date, the vast majority of research into telerobotics has focused on enabling normal people to operate a robot safely and efficiently in extreme work environments such as deep oceans, space, battlefield, etc. This project focuses on the opposite end of the telerobotics spectrum; how to allow special people (those with physical disabilities) to operate a robot safely and comfortably in normal, routine work environments such as police stations, fire departments, hospitals, construction sites, business offices, etc. We believe that the importance of this research is to develop general purpose telepresence robots that are affordable, reliable, sociable public appearance, and human-centered interface for disabled teleoperators to intuitively grasp and use. In FIU Discovery Lab, we have been developing a low cost telepresence humanoid robot, known as TeleBot that would enable a disabled veteran or police officer to perform law enforcement duties. The project team members studied the design of two military-grade robots that were borrowed from Florida Institute of Human Machine Cognition (IHMC) at Pensacola, FL. Subsequently, the team has developed the first TeleBot prototype with video, audio and sensory capabilities so that a disabled teleoperator can intuitively control the TeleBot. I will introduce architectural and algorithm designs of this project for the realization of real-time telepresence operations with optimum bandwidth utilization, human-centered intuitive control interface and ergonomic movements as well as motivation and development process of this project. Dr. Jong-Hoon Kim is the Director of FIU Discovery Lab and a Visiting Assistant Professor in the School of Computing and Information Sciences, Florida International University. He received a B.S. from Seoul National University of Science and Technology, Seoul, South Korea in 2005. He received his masters and Ph.D. in the Department of Computer Science, Louisiana State University at Baton Rouge in 2008 and in 2011 respectively. His research interests are in Mobile Robots, Telerobotics, Embedded Systems, HCI, Sensor Networks, and Intelligent Systems. Currently, his research is focusing on Tele-Presence Robot (TeleBot), and Intelligent System for Home Automation (ISHA). This is another in the Departmental Pizza Seminar series. #### Previous Colloquia ## Previous Colloquia Announcements Wednesday 4th December 2013, 3:00pm, UB411 (pizza @ 4:30pm), UB230 (talks @ 3:00pm and 5pm) Computer Science Undergraduate Students will present Project and Internship Presentations ### Jiajun Chen - Analyzing GRIB Data ### Sean Meadows - Discovering Fate ### Daniel Rangel - Data Analysis on Theorem Proving Systems ### Matt Sanchez - Bitcoin Mining ### Jack Bobrow - Basketball Simulator ### Alyssa Hewitt - LPS Group Internship ### Matt Ragonese - JogLog This is another in the Departmental Pizza Seminar series. Wednesday 20th November 2013, 5:00pm, UB411 (pizza @ 4:30pm), UB230 (talk @ 5:00pm) Sawsan Khuri will present Cross-Cultural Collaboration in Scientific Discovery or "How to Talk to Each Other" Scientific discovery today is undertaken for the most part by teams working together in a cross-disciplinary setting.Further afield and wider interactions are being encouraged by funding agencies and university administration, but academics and students who are expected to work in these disparate teams are often ill equipped to cope with the dynamics of setting up cross-disciplinary teams and of making them successful. Often in the biomedical sciences, teams include molecular biologists or biochemists working together with computer scientists or engineers -- disciplines that have different languages and different cultures. Traditional boundary lines between these disciplines are becoming blurred, and there are several new disciplines that have emerged along that continuum, such as bioinformatics. This talk will introduce the topic of Team Science, discuss the ways in which collaborating across disciplines is in fact a crossing of cultures, and give examples of the joys and woes of being team members and leaders of cross-disciplinary collaborative projects. This is another in the Departmental Pizza Seminar series. Wednesday 13th November 2013, 5:00pm, UB411 (pizza @ 4:30pm), UB230 (talk @ 5:00pm) Eddy Borja will present Computer Science and the Entrepreneurial Revolution Computer Science and Entrepreneurship have long had a close relationship. The advantages of software based businesses over traditional ones can empower an individual to accomplish huge goals with minimal starting resources. Why pursue this path now? There has never been a better time to get started: the time and costs of building a new business or startup are falling dramatically. In recent years, new methodologies such as the Lean Startup movement have developed risk-reducing practices for developing businesses and products through a process now called 'validated learning'. Community building events like Startup Weekend are happening globally and bringing experienced based education to people interested in entrepreneurship, but unsure of how to take the first step. Through these kind of events and more, people of different backgrounds are networking together to forge great ideas into products and services that could change your world. Even with this however, the world still needs more entrepreneurs if the promises of a better future are to someday be held true. Stay hungry, stay foolish! This is another in the Departmental Pizza Seminar series. Wednesday 6th November 2013, 5:00pm, UB411 (pizza @ 4:30pm), UB230 (talk @ 5:00pm) Amy Eguchi Bloomfield College will present Robotics in Education from K to 12 and More Robotics in Education has been one of the most talked about topics of technology in education for several years. Mataric, a professor of USC and the director of Center for Robotics and Embedded Systems and Robotics Research Lab, stated that "robotics has been a growing field with the potential to significantly impact the nature of engineering and science education at all levels, from K-12 to graduate school." The talk introduces the Robotics in Education in K-12 classroom, and how using the educational robotics as a learning tool helps to enhance students learning experience of subject content knowledge, especially in STEM fields. In addition, it talks about how educational robotics can enhance the learning of 21st century skills, design thinking, and engineering thinking skills, which are some of the most important skills that 21st century students need to be successful innovators. Amy Eguchi is an Associate Professor of Education at Bloomfield College in New Jersey. She holds her M.A. in Child Development from Pacific Oaks College, Ed.M. in Education from Harvard Graduate School of Education, and Ph.D. in Education from the University of Cambridge and has an extensive teaching experience in educational robotics both with students and teachers in K-12 setting. She also teaches educational robotics to undergraduates. She runs a competitive robotics after school team at The School at Columbia University. Dr. Eguchi has been involved in RoboCupJunior, an educational robotics competition, since 2000, as the technical committee and organizing committee members, as well as the co-chair and general chair, in international, national, and local levels. In addition, she is the vice president of RoboCup Federation representing RoboCupJunior, and a member of the RoboCup Federation Board of Trustees. Dr. Eguchi has been involved in several international collaboration educational robotics projects including the CoSpace educational robotics projects with the Advanced Robotics and Intelligent Control Centre (ARICC) at Singapore Polytechnic, Singapore and RoboCupJunior initiative in Bangladesh. She also provides consultations to various educational robotics initiatives from around the world. This is another in the Departmental Pizza Seminar series. Wednesday 30th October 2013, 5:00pm, UB411 (pizza @ 4:30pm), UB230 (talk @ 5:00pm) Hande Küçük-McGinty Department of Computer Science University or Miami will present Knowledge modeling and integration of assays and data from the Library of Integrated Network-based Cellular Signatures (LINCS) The National Institute of Health (NIH) Library of Integrated Network-based Cellular Signatures (LINCS) program is generating extensive multidimensional datasets including biochemical-, genome-wide transcriptional-, and phenotypic cellular response signatures to a variety of small molecule and genetic perturbations with the goal to create a sustainable, widely applicable, and readily accessible systems biology knowledge resource. Integration and analysis of diverse LINCS datasets could be better accommodated by using semantic web technologies. We introduce the knowledge model developed and implemented for the LINCS data, how we implemented a complex systems biology model using OWL DL, and how it might help life scientists better analyze the datasets by using the reasoning capabilities. We further discuss how the analysis could be improved by integrating additional data, such as biological pathways by using semantic web technologies. The discussion will continue on to show how we can utilize the experiments that measure cellular pathway information by using different time points and how this information can be used for better analysis by using temporal reasoning. This is another in the Departmental Pizza Seminar series. Wednesday 23rd October 2013, 5:00pm, UB411 (pizza @ 4:30pm), UB230 (talk @ 5:00pm) Yajie Hu Department of Computer Science University or Miami will present A Model-based Music Recommendation System for Individual Users and Implicit User Groups This talk introduces several relevant problems on music recommendation. To make a high-qualified recommender, we evaluate feature importance for favorite song detection from two perspectives. The experiment results expose that collaborative filtering signal is the most important feature among the selected features. A classifier combination method is proposed to leverage several classifiers trained by different data sources to predict music genre. The complemented genre labels are used in a recommendation system for individual users on local device. The recommender takes freshness, time pattern, genre, publish year, and favor into account to make recommendations. The recommender outperforms the baseline on mostly favorite songs. Furthermore, a probability-based method is proposed to make recommendations for implicit user groups by integrating individual opinions on candidate songs. At the end, a possible adapted recommendation method is introduced to response user feedback and find out local optimizations to improve the recommendation quality. This is another in the Departmental Pizza Seminar series. Wednesday 9th October 2013, 5:00pm, UB411 (pizza @ 4:30pm), UB230 (talk @ 5:00pm) Justin Stoecker Department of Computer Science University or Miami will present Interactive Visualization Systems and Sensor-Based Interfaces for Robotics and Medical Data Analysis Scientists tend to collect lots of data, and the problem arises: how does one interpret all of this information? This talk will cover scientific visualization projects in the domains of robotics and medicine. These research efforts leverage the modern graphics rendering pipeline and accessible sensors (Microsoft Kinect and Leap Motion) to provide interactive experiences that enable scientists and doctors to extract insight from their data. The first part of this talk will focus on recent work: visualizing robot behaviors in multi-agent environments, modeling robot configuration spaces, and mapping human motions to humanoid robots. The main part of this talk covers a collaboration with doctors at the Oregon Health & Science University Hospital; we are working on an interactive gesture-controlled visualization system for intraoperatively viewing and manipulating data in a sterile environment. This is another in the Departmental Pizza Seminar series. Wednesday 2nd October 2013, 5:00pm, UB411 (pizza @ 4:30pm), UB230 (talk @ 5:00pm) Piyali Nath Department of Computer Science University or Miami will present Distributed/Parallel Optimization of Captured Human Motions for Generating Robust and Stable Motions on Simulated Robots Motion learning is an interesting topic for research. The current approach used for motion learning in the RoboCanes framework uses the process of recording a human motion and optimizing the same to get a stable motion for a simulated robot. But the major downside of this approach is that this optimization process takes hours to return a good set of parameters. One possible solution is to parallelize the optimization process. This talk is about an extension of the existing framework to generate stable motions for simulated humanoid robots using motion capture framework and then optimizing them efficiently in a distributed/parallel environment. This approach uses a master slave network of systems where the master system controls the optimization process and distributes the particles (candidate solutions of an optimization process) to multiple client systems which are responsible for running simulations individually, using the parameters sent by the server and then evaluating the error and returning it to the server. The main objective is to enable faster iterations and thus reduce the time for finding the best value for the set of parameters used in the fitting function for all the joints of the motion. This is another in the Departmental Pizza Seminar series. Wednesday 25th September 2013, 5:00pm, UB411 (pizza @ 4:30pm), UB230 (talk @ 5:00pm) Chaoming Song Department of Physics University or Miami will present Understanding Complex Systems: Networks and Dynamics Fueled by a wealth of data supplied by a wide range of high-throughput tools and technologies, the study of complex systems is currently reshaping a number of research fields from social science to computer science and biology. This data-rich reality calls for new approaches and techniques to harvest the hidden information and devise new models to explain the underlying principles of various complex systems. While from a functional standpoint different systems may appear to be distinct from one another, there is an increasing realization that they often share similar structural and dynamic properties. Such similarities offer new perspectives and unique opportunities for physicists to apply their methodologies on a much broader set of phenomena. In this talk, I will first present a macroscopic study of large-scale network structures observed in diverse datasets, and next focus on understanding social activities such as communication and traveling pattern at the each individual level. In the end, I will show a series of relationships that link the quantities characterizing social networks and human dynamics, and demonstrate their generality across a wide range of systems, from mobile calls to tweets. This is another in the Departmental Pizza Seminar series. Wednesday 11th September 2013, 5:00pm, UB411 (pizza @ 4:30pm), UB230 (talk @ 5:00pm) Geoff Sutcliffe Department of Computer Science University or Miami will present How to Give a Successful Talk Almost everyone has to give a public presentation at some time. Graduate students have to defend their theses, research students have to present their results, and many jobs require presentations. It is common to have little or no experience when you give your first presentation, and you may even be a little nervous! This talk is aimed at (graduate, research, and other) students. It describes how to present a successful talk, in a simple standard format. This talk does not try to teach general speaking skills, nor impose any personally preferred techniques. It covers the structure of a talk, the use of visual aids, speaking technique, and how to cope with questions. This is another in the Departmental Pizza Seminar series. Wednesday 4th September 2013, 5:00pm, Room 411 (pizza @ 4:30pm), 230 (talk @ 5:00pm) Magnolia will present The Magnolia Content Management System Today's web is more complex than ever before. Variations in devices, platforms, and channels makes delivering content correctly a tedious job. For large, fast-paced companies, keeping your web presence up-to-date with the latest content can be challenging. Integration with legacy systems and data storage is a must but adds multiple layers of complexity. Magnolia CMS makes complex websites simple to manage, while its flexible software architecture makes it possible to implement custom functionality. During this presentation Magnolia will provide an overview of its Content Management System (CMS) technology. The presenters will discuss why a CMS is needed to manage today's enterprise websites. The presentation will touch on the core technologies used by the system, a high level overview of how content storage is handled, and the modular nature of the Magnolia CMS product. Attendees should walk away from this presentation with a better understanding of how an Enterprise-level Java system is built. This is another in the Departmental Pizza Seminar series. Wednesday 14th April 2013, 5:00pm, Room 411 (pizza @ 4:30pm), 311 (talk @ 5:00pm) ## Spring 2013 Undergraduate Project Presentations Jiabin Fu will present Solving Computational Complexity Problem by Playing 17x17 Game Micah Reback will present Opinion Spam Detection Through Data Mining This is another in the Departmental Pizza Seminar series. Wednesday 10th April 2013, 5:00pm, Room 411 (pizza @ 4:30pm), 311 (talk @ 5:00pm) Andreas Seekircher Department of Computer Science University of Miami will present Stable and Flexible Walking Motions for Physical Humanoid Robots Controlling an autonomous, humanoid robot in a real time, dynamic, and adversarial environment is a complex task that includes several areas of computer science. I will start with an overview of controlling a robot and present some general concepts. RoboCanes, UM's robotic soccer team, participated during the last few years in the RoboCup, an international robotic soccer competition fostering research in both robotics and AI. RoboCup uses soccer as a standard problem that provides many opportunities to develop general methods for controlling robots in demanding environments. An essential ability of a robot is to act in its environment by generating motions. This can be a challenging problem on a robot with > 20 degrees of freedom. The current solutions in the RoboCanes agent include optimization methods that work well in a simulated environment. However, a stable bipedal walk for a physical robot requires additional approaches. One common approach is to generate trajectories of the zero moment point and the center of mass under constraints given by a simplified physical model such as an inverted pendulum. A lot of research has been done in this area in the recent years. However, despite good results in theory and simulations, in the end most implementations on physical robots still need a lot of manual tuning and are not very robust with respect to changes in the environment. Possible reasons for that and ideas for improvement will be discussed. Andreas Seekircher is a PhD student at the University of Miami. He holds an MSc in Computer Science from the University of Bremen, Germany. His research is focused on robotics and artificial intelligence, specifically controlling autonomous robots. This is another in the Departmental Pizza Seminar series. Thursday 25th April 2013, 3:30pm, Room 311 ## Spring 2013 Undergraduate Project Presentations Isabella Douzoglou will present A Simple Approach to More Accurately Recommended Music Lauren Ramgattie will present Booz Allen Hamilton Summer Internship: Mobile Security Research Analyst This is another in the Departmental Pizza Seminar series. Wednesday 3rd April 2013, 5:00pm, Room 411 (pizza @ 4:30pm), 311 (talk @ 5:00pm) Charlotte Szostek Bristol Centre for Complexity Sciences will present Equation-free Stability Analysis of Trading Agents Acting on Different Timescales I will discuss how equation-free methods can be used to describe the dynamics and quantify stability of an Agent Based Model where automated trading agents interact in a Continuous Double Auction. Agents are heterogeneous in their decision making and in the timescales at which they receive, process and respond to market information. The stability of asset price and proxy indicators of order book dynamics is compared for different timescale variations. Such dynamics have not as yet been analytically quantified although recent work strongly suggests that timescales are an important component. Modern day financial markets are composed of trading agents interacting on a multitude of timescales. Depending on the market, up to 70% of trading may be due to automated algorithmic trading. These agents make and execute decisions on millisecond time scales, whilst human traders are restricted by their reaction times---in the order of 1000 milliseconds. Statistical evidence (/Johnson et al. 2012/) and experimental evidence /(Cartlidge & Cliff, 2012 and 2013, Cartlidge, De Luca, Cliff & Szostek, Forthcoming) /suggests that there exists, in effect, 'Parallel markets' that collectively determine//the order book dynamics of an asset. Automated-automated interactions on//microsecond timescales, human-human interactions occur at a timescale of//seconds, and depending on the time-scale difference between automated and//human agents; a varying quantity of automated-human interactions.//Experimental evidence (/Cartlidge, De Luca, Szostek, & Cliff, (2012)) /also shows//timescales impacts mechanisms that determine order book dynamics: specifically//the efficiency of convergence of trader shouts to the 'fundamental' market price. Charlotte Szostek is a PhD candidate at the Bristol Centre for Complexity Sciences in the UK. Her research interests are in Non--Linear Dynamics and the use of Equation-Free methods to perform stability analysis, specifically on ABMs. The application areas of her research are financial markets and automated trading. For the past six months Charlotte has been investigating the feasibility of ABMs to inform risk more accurately for hedge funds as agents in the financial system and to understand systemic risk due to their behavior. For this time she has been based at a Risk Services Provider in New York. Charlotte's publications to date have been 'experimental economics' in computer science journals - observing interactions between human traders and automated algorithms. She has contributed to the UK governments Foresight report on the 'Future The Future of Computer Trading in Financial Markets, Driver Reviews. This is another in the Departmental Pizza Seminar series. Wednesday 27th March 2013, 3:30pm, Room 411 Dr. Steffen Wachenfeld PhD, University of Münster, Germany Co-Founder of Capino GmbH will present Recognition of Text from Screenshot Images Handwriting recognition works quite well today and optical character recognition (OCR) for scanned documents is believed to be nearly error free. In this talk, the recognition of text from screenshot images will be presented which comes with very strong challenges. The biggest challenge is the very low resolution of rendered characters which typically have a height of just a few pixels. The talk will start with a brief introduction to typography and rendering technologies related to the creation of screen-rendered text, followed by a motivation for the need for text recognition from screenshot images and the presentation of selected aspects of a recognition system that was implemented at the University of M?nster, Germany. The implemented system was tested on publicly available databases holding thousands of screenshot images of single characters and words. To the author's knowledge, the system is currently the best performing system on screen-rendered text including state-of-the-art text recognition programs. Dr. Steffen Wachenfeld received his PhD in Computer Science from the University of M?nster, Germany. He was conducting research and teaching at the Department of Computer Science for six years. Besides his PhD he holds a Masters Degree in Computer Science and a Masters Degree in Information Systems. His fields of expertise comprise social media, mobile computing, image analysis, pattern recognition, artificial intelligence, networking, mobile computing, and databases. After graduating, Steffen worked for two years in Philadelphia, USA, as a postdoctoral researcher at the University of Pennsylvania in the field of Biomedical Image Analysis. Besides his academic career, Steffen ran an IT-related business for over 10 years and was co-founder of Capino GmbH, a company developing language learning software. Also he spent two years at The Boston Consulting Group, one of the world's leading Consulting Companies. This is another in the Departmental Pizza Seminar series. 1st March 2013, 3:30pm, Room 411 Ziya Arnavut Department of Computer and Information Sciences State University of New York, Fredonia will present A Dynamic Pseudo-Distance Technique with Context-Modeling for Lossless Compression on Color-Mapped Images This talk introduces a dynamic pseudo-distance technique with a context-modeling for compression of color-mapped images. Graphic Interchange Format (GIF) and Portable Network Graphics (PNG) are two of the well-known and frequently used techniques for the compression of color-mapped images. There are several techniques that achieve better compression results than GIF and PNG; however, most of these techniques need two passes on the image data, while others do not run in linear time. The pseudo-distance technique runs in linear time and requires only one pass. In this talk, we show that with a modified pseudo-distance technique and context-modeling, further compression gains are possible over both GIF and PNG. This is a Department of Computer Science Colloquium. Wednesday 20th February 2013, 5:00pm, Room 411 (pizza @ 4:30pm), 311 (talk @ 5:00pm) Alexander Härtl Department of Computer Science University of Miami will present Robust Real Time Object Recognition for a Humanoid Soccer Robot In humanoid robot soccer, the color coded environment is predominantly perceived through a camera sensor. Therefore, a reliable and efficient object recognition system is crucial. Over the years, a standard approach evolved that uses a static color classification as the first processing step. Besides its efficiency, this approach lacks robustness against unforeseen changes in illumination. In this talk, I will present an object recognition system for the RoboCup Standard Platform League that does not use static color classification as the first processing step. Instead, objects are found based on color similarities to increase robustness. The proposed system is still sufficiently efficient to allow for real time operation despite the limited processing power of the robot. It is shown that the detection rate is comparable under static lighting conditions and substantially better under changing lighting conditions. This is another in the Departmental Pizza Seminar series. Wednesday 13th February 2013, 5:00pm, Room 411 (pizza @ 4:30pm), 311 (talk @ 5:00pm) Milica Mormann Senior Research Associate University of Miami will present Predicting Visual Attention and Exploring Its Influence on Choices We are exposed to enormous amounts of information, yet our processing capacity is limited. How we attend to and perceive incoming information (e.g., snack food options) has a profound influence on our choices (e.g., chosen snack item). In particular, choices are affected by exogenous factors such as visual saliency, which affects where we look and thus which alternatives we consider, and endogenous factors such as expected reward, which also affects our eye-movements and choices. A biologically plausible model of early visual attention from computational neuroscience is used to analyze the physical characteristics of objects in a visual scene in order to determine which items are most likely to attract viewers' attention. At the same time, a series of eye-tracking experiments examine where attention is deployed during real decision-making process. Finally, a computational model (i.e., a drift-diffusion type of model) is introduced, which describes (and aims to predict) how visual and reward information determine economic choices. This is another in the Departmental Pizza Seminar series. Wednesday 6th February 2013, 5:00pm, Room 411 (pizza @ 4:30pm), 311 (talk @ 5:00pm) Saminda Abeyruwan Department of Computer Science University of Miami will present Learnable Knowledge Representations for Autonomous Agents With the advancement of computational infrastructure and the success of the statistical machine learning methods, there exists a substantial problem of interpreting the content. For an autonomous agent: what is knowledge? Is it learnable? What constituents keep the knowledge correct? And: where does an agent get the ground truth from to verify knowledge? High-level languages or methods such as description logic (SROIQ(D)) or Bayesian networks are the most prevalent technologies to represent knowledge with substantial expressivity and moderate computational cost. These representations allow an agent to infer and reason about static knowledge. There remains the question whether an agent can dynamically predict knowledge, learn knowledge from sensorimotor interactions and generalize for unseen circumstances. In this talk we look at a mechanistic solution for these questions using description logic representation and reasoning subsystem to synchronize high-level human knowledge (essentially public) with the low-level sensorimotor knowledge (essentially private). This is another in the Departmental Pizza Seminar series. Monday 28th January 2013, 11:00am, Executive Conference Room, Newman Alumni Center Stefan Wuchty National Center of Biotechnology Information National Institutes of Health will present Computational Biology in Infectious Diseases and Cancer The abundance of many disparate biological data sources necessitates the development of computational methods that allow skillful data integration. Specifically, such approaches provide the opportunities to identify complex relationships between biological entities that can shed a light on key players in diseases. Giving an overview of applications of novel approaches I will present (i) a current flow algorithm that allows the determination of causal paths through molecular interaction networks from genomic perturbations to disease genes in brain cancers. (ii) Furthermore, I will introduce a graph-based approach that was applied to determine epistatic interactions between genes in the human malaria parasite. (iii) Finally, I will present a simple computational approach to predict microRNAs that potentially play a role in different cancers. This is a College of Arts & Sciences and Department of Computer Science colloquium. Wednesday 28th February 2013, 5:00pm, Room 411 (pizza @ 4:30pm), 311 (talk @ 5:00pm) Joseph Clarke Distinguished Services Engineer Cisco will present Software Defined Networking There is a lot of buzz in the networking industry these days about Software Defined Networking or SDN.Cisco recently announced a strategy around SDN and network programmability called the Open Network Environment or ONE.One of the key pillars of ONE is a rich device-level application programming interface (API) called onePK or ONE Platform Kit. This talk will focus on what onePK provides in terms of APIs cand capabilities.Attendees will have a chance to think in new terms about how networks can be provisioned, managed, and controlled.Some use case demonstrations will be shown to help reinforce the material. This is another in the Departmental Pizza Seminar series. Wednesday 23rd January 2013, 3:30pm, McArthur Engineering Annex, Room 202 Steven M. Watt University of Western Ontario will present The Mathematics of Mathematical Handwriting Recognition Accurate computer recognition of handwritten mathematics offers to provide a versatile interface for mathematical computing systems, such as Maple, document creation and collaboration. Mathematical handwriting, however, provides a number of challenges beyond what is required for the recognition of handwritten natural languages. New methods of character recognition are important for accurate handwritten mathematics input. We present a geometric theory that we have found useful for recognizing mathematical symbols. Characters are represented as parametric curves approximated by certain truncated orthogonal series. This maps symbols to a low-dimensional vector space of series coefficients in which the Euclidean distance is closely related to the variational integral between two curves. This can be used to find similar symbols very efficiently. We describe some properties of mathematical handwriting data sets when mapped into this space and compare classification methods and their confidence measures. The beauty of this theory is that a single, coherent view provides several related geometric techniques that give a high recognition rate and that do not rely on peculiarities of the symbol set. This is another in the Departmental Pizza Seminar series. Thursday 17th January 2013, 3:30pm, Executive Conference Room, Newman Alumni Center Bonnie Kirkpatrick Department of Computer Science University of British Columbia will present Biological Models with Combinatorial Spaces Probabilistic models are common in biology. Many of the successful models have been readily tractable, leaving calculations on models with a combinatorial-sized state space as an open problem. This talk examines two kinds of models with combinatorial state spaces: continuous-time and discrete-time Markov chains. These models are applied to two problems: RNA folding pathways and family genetics. While the applications are disparate topics in biology, they are related via their models, the statistical quantities of interest, and in some cases the computational techniques used to calculate those quantities. This is a Department of Computer Science colloquium. Wednesday 14th November 2012, 5:00pm, Ungar Building, Room 411 (pizza), 311 (talk) Dr Sawsan Khuri Center for Computational Sciences University of Miami will present Team Science and Bioinformatics Look at any good medical science research article in any high impact journal. Count the number of authors and look at the range of departments and institutions they come from. Chances are that somewhere in the authors list are one or more bioinformatics experts without whom the project could not have been completed. Research in today's world is about teams of people from an increasing number of disciplines, and in the medical sciences these teams include clinicians, biochemists, molecular biologists, bioinformaticians, statisticians, and software engineers. This talk will highlight the rise of team science, the launch of the Science of Team Science field, and their main findings about how scientific teams work (or don't!). Examples will be presented of successful team science projects that include bioinformatics, such as the Human Genome Project and from the drug discovery world. This is another in the Departmental Pizza Seminar series. Wednesday 7th November 2012, 5:00pm, Ungar Building, Room 411 (pizza), 311 (talk) Justin Stoecker Department of Computer Science University of Miami will present Visualization and Path Planning with 3D Configuration Spaces Consider a 2D scene with a static obstacle, such as a top-down view of an apartment, and a mobile robot capable of 2D translation and 1D rotation. The robot can move with three degrees of freedom, but it cannot collide with the obstacle. The space of all valid arrangements of the robot is called its configuration space; it consists of all (X,Y,?) such that the robot is not overlapping with the obstacle. This talk will illustrate how the configuration space of such a scene can be visualized by constructing and rendering a 3D polygonal mesh that exists at the configuration space boundaries. The mesh is built "airtight" and without fixed sampling along ? in order to attain a compact and accurate representation of the smooth surface of the configuration space. The triangles of the mesh can be used to generate a path for the robot from an initial configuration to a goal configuration without collision. This is another in the Departmental Pizza Seminar series. Monday 14th January 2013, 11:00am, CAS Wesley Gallery Tiago Branco will present Dendritic Information Processing in Cortical Pyramidal Cells Information is delivered to cortical pyramidal neurons via thousands of synaptic inputs, activated at different dendritic locations with varying degrees of temporal synchrony and in different sequences. Using electrophysiological recordings combined with two-photon glutamate uncaging, calcium imaging and compartmental modeling, we tested how different regions along single cortical dendrites integrate excitatory inputs, and whether they are sensitive to the sequence of synaptic activation. We found that dendrites of cortical neurons have a gradient of non-linear synaptic integration, whereby proximal inputs sum linearly and require precise temporal coincidence for effective summation, whereas distal inputs are amplified with high gain and integrated over broader time windows. The mechanism involves dendritic impedance gradients and non-linear synaptic NMDA receptor activation, and also confers high sensitivity to the temporal input sequence, allowing dendrites to efficiently discriminate different sequences of synaptic activation. Pyramidal cell dendrites can thus exhibit multiple computational strategies, and act processing compartments for detection of synaptic sequences, implementing a fundamental cortical computation. Tiago Branco received his M.D. from Lisbon University in 2002. He then joined the Wellcome Trust Four-Year Ph.D. Programme in Neuroscience at University College London (UCL), where in the group of Dr. Yukiko Goda he focussed on neurotransmitter release properties of individual synapses. After receiving his Ph.D. in 2007 he moved to Prof. Michael Hausser's laboratory in UCL, where he applied electrophysiological, optical and modelling techniques to investigate how dendritic integration contributes to single neuron computations. In 2012 he started his own group as a Sir Henry Dale Fellow at the MRC Laboratory of Molecular Biology, where he combines physiological and molecular methods to investigate the how the mouse brain implements the computations that underlie innate behaviours. This is a Departmentof Computer Science colloquium. Wednesday 31st October 2012, 5:00pm, Ungar Building, Room 411 (pizza), 311 (talk) Changwon Yoo, S.M., Ph.D. Department of Biostatistics Florida International University will present Identifying Gene Alterations Required for the Development of Astrocytoma Our goal in this study was to identify gene-environment interactions that are critical in the development of astrocytic tumors. We also aim to identify genes and/or pathways that may further the understanding of the differences between low and high grade astrocytic tumors. Several steps were involved in our analysis, including: 1) identification of a significant set of over-and under-expressed genes through meta-analysis of several astrocytoma microarray studies; 2) enrichment analysis of the set of significant genes; 3) network analysis of the set of significant genes; and 4) investigation and validation of the network analysis. Through meta-analysis of 12 sub-studies which compared normal tissue to astrocytomas, we were able to identify a list of 554 genes which were differentially expressed in the majority of these studies. Many of the genes identified in this study have in fact been implicated in the development of astrocytoma, including EGFR, HIF-1α, c-Myc, WNT5A, and IDH3A. We then performed reverse engineering of our gene list using Bayesian network analysis. Networks of genes were produced for astrocytoma. This study was able to identify a set of key genes significantly disregulated during the development of astrocytoma. Our results suggest that alterations in the expression of eight to ten key genes may be required for the development of astrocytoma. This is another in the Departmental Pizza Seminar series. Monday 22nd October 2012, 11:00am, CAS Wesley Gallery Raissa D'Souza Department of Computer Science Department of Mechanical and Aerospace Engineering Complexity Sciences Center University of California, Davis will present Percolation, Cascades, and Optimal Interdependence of Networks Collections of networks are at the core of modern society, spanning technological, biological and social systems. Over the past decade, a science of networks has been emerging and providing insights into the structure and function of many diverse types of systems, such as protein-interactions in a cell, collaboration networks of scientists, and the World Wide Web. Random graphs provide a framework for modeling network phenomena, especially phase transitions, such as the sudden emergence of large-scale connectivity. This talk will give an overview and present a variant of the classic Erdos-Renyi model of network formation, showing that we can alter the location and also the nature of the phase transition, making for an explosive onset of connectivity. We also develop random graph models of interacting networks, motivated by the fact that individual networks are increasingly interdependent (e.g., the Internet and the power grid, globalization of financial markets). We show that interactions between different types of networks can actually lower critical thresholds and provide stabilizing effects with respect to cascades. Raissa D'Souza is an Associate Professor of Computer Science and of Mechanical Engineering at the University of California, Davis, as well as an External Professor at the Santa Fe Institute. Her research focuses on mathematical models of self-organization, phase transitions, and the structure and function of networked systems. Raissa received a PhD in Statistical Physics from MIT in 1999, then was a postdoctoral fellow, first in Fundamental Mathematics and Theoretical Physics at Bell Laboratories, and then in the Theory Group at Microsoft Research. Her influential, interdisciplinary work on network theory has appeared in journals such as Science, PNAS, and Physical Review Letters, and her research has been funded by the NSF, the National Academy of Sciences of the USA, DOD, ARL and AFOSR. Raissa has been selected a Kavli Fellow of the National Academy of Sciences four times and is a member of the Science Steering Committee of the Santa Fe Institute. She has served on the program committees for numerous conferences, and currently serves on the editorial board of several international mathematics and physics journals. When not pursuing research ideas, she can typically be found outside, either scaling rocks or sailing. This is a College or Arts and Sciences, and Department of Computer Science, Colloquium. Wednesday 17th October 2012, 5:00pm, Ungar Building, Room 311/411 Negin Arhami Department of Computer Science University or Miami will present Strong Logic and Weak Reasoning vs. Weak Logic and Strong Reasoning There are different forms of logic with different levels of expressivity. Scientists have developed Automated Theorem Provers (ATPs) to reason about these different forms of logic. An ATP is a tool that automatically checks if a statement (conjecture) is logical consequence of a set of statements (axioms). Very expressive logics have less powerful ATPs, and conversely less expressive logics have more powerful ATPs. A problem, which is a set of statements in a natural language, is first stated in a logical form. The logical form is either given to a suitable ATP, or can be translated to a less expressive logic with more powerful ATPs. The question is which approach gives the best trade-off between expressivity and ATP performance. This is another in the Departmental Pizza Seminar series. Wednesday 10th October 2012, 5:00pm, Ungar Building, Room 311/411 Prof. Victor Milenkovic Department of Computer Science University or Miami will present Robust Complete Path Planning in the Plane We present a complete path planning algorithm for a plane robot with three degrees of freedom and a static obstacle. The part boundaries consist of n linear and circular edges. The algorithm constructs and searches a combinatorial representation of the robot free space. Its computational complexity is O((n4 + c3 )log n) with c3 O(n6 ) the number of configurations with three simultaneous contacts between robot and obstacle edges. The algorithm is implemented robustly using our adaptive-precision controlled perturbation library. The program is fast and memory efficient, is provably accurate, and handles degenerate input. Joint work with Elisha Sacks of Purdue University and Steven Trac of Epic Systems Corporation. This is another in the Departmental Pizza Seminar series. Wednesday 12th September 2012, 5:00pm, Ungar Building, Room 506 Geoff Sutcliffe Department of Computer Science University or Miami will present How to Give a Successful Talk Almost everyone has to give a public presentation at some time. Graduate students have to defend their theses, research students have to present their results, and many jobs require presentations. It is common to have little or no experience when you give your first presentation, and you may even be a little nervous! This talk is aimed at (graduate, research, and other) students. It describes how to present a successful talk, in a simple standard format. This talk does not try to teach general speaking skills, nor impose any personally preferred techniques. It covers the structure of a talk, the use of visual aids, speaking technique, and how to cope with questions. This is another in the Departmental Pizza Seminar series. Wednesday 5th September 2012, 5:00pm, Ungar Building, Room 506 Walid Saad Department of Electrical and Computer Engineering University or Miami will present Toward Smart, Secure, and Self-Organizing Wireless Networks Next-generation wireless systems are characterized by three key features: heterogeneity, in terms of technologies and services; size, in terms of number of users, nodes, and services; and dynamics, in terms of rapidly varying environments and uncertainty. The design and analysis of such complex systems must handle a myriad of co-existing devices that can interact, self-organize, meet stringent QoS requirements, and adapt to their environment. In this respect, game theory is expected to play a critical role toward deploying intelligent, distributed, and self-organizing communication systems in which network devices can make independent and rational strategic decisions. This talk provides a comprehensive introduction to game theory in its two branches, non-cooperative and cooperative games, as it applies to the design of future heterogeneous wireless networks. For each type of games, we present the fundamental components, introduce the key properties, mathematical techniques, solution concepts, and describe the challenges and methods for applying these games in two emerging fields: (i) the design and analysis of heterogeneous small cell and cognitive wireless networks and (ii) the development of physical layer wireless security solutions. We conclude the talk with insights on future work and some related areas. Walid Saad received his B.E. from the Lebanese University in 2004, his M.E. in Computer and Communications Engineering from the American University of Beirut in 2007, and his PhD degree from the University of Oslo in 2010. From August 2008 until July 2009 he was a visiting scholar in the Coordinated Science Laboratory at the University of Illinois at Urbana Champaign. From January 2011 until July 2011, he was a Postdoctoral Research Associate at the Electrical Engineering Department at Princeton University. Currently, he is an Assistant Professor at the Electrical and Computer Engineering Department at the University of Miami. His research interests span the areas of game theory, wireless networks, small cell networks, cognitive radio, and security. He has published over 60 publications in major technical journals and conferences in these areas, including a book on game theory in wireless and communication networks, published by Cambridge University Press in October 2011. He has recently received three best paper awards at the 7^th International Symposium on Modeling and Optimization in Mobile, Ad Hoc and Wireless Networks (WiOpt), in June 2009, at the 5^th International Conference on Internet Monitoring and Protection (ICIMP) in May 2010, and at the IEEE Wireless Communications and Networking Conference (WCNC) in 2012. This is another in the Departmental Pizza Seminar series. Thursday 9th August 2012, 11:30am, Ungar Building, Room 402 Prof. Erol Gelenbe Professor in the Dennis Gabor Chair, Imperial College Member French Nat'l Academy of Engineering (Academie des Technologies) Fellow of IEEE (1986) and ACM (2001) Head of Intelligent Systems and Networks, Dept. Electrical & Electronic Engineering will present Networks of Networks in the Digital Society Our digital society relies on commerce, work, food provisioning, transportation, energy, information and data networks, and many others, that interact to make the system "work as a whole". For the first time in human history, all of these networks have a common mediator other than money, and a representation system: the information and data networks. Thus we can envision to model and understand these complex networks as an observable formal system. This talk will present an approach to such "NetONets" based on stochastic networks that leads to closed form mathematical solutions, and we will illustrate it through specific results for stochastic networks that simultaneously model commerce, transportation, ICT and energy consumption. This is a Department of Computer Science Colloquium. Refreshments will be served at 11:00am Monday 11th June 2012, 11:30am, Ungar Building, Room 426 Dr. Evangelos Markatos Department of Computer Science University of Crete will present Privacy-Preserving Social Plugins The widespread adoption of social plugins, such as Facebook's Like and Google's +1 buttons, has raised concerns about their implications to user privacy, as they enable social networking services to track a growing part of their members' browsing activity. Existing mitigations in the form of browser extensions can prevent social plugins from tracking user visits, but inevitably disable any kind of content personalization, ruining the user experience. In this work we describe a novel design for privacy-preserving social plugins that decouples the retrieval of user-specific content from the loading of a social plugin. In contrast to existing solutions, this design preserves the functionality of existing social plugins by delivering the same personalized content, while it protects user privacy by avoiding the transmission of user-identifying information We have implemented our design in SafeButton, an add-on for Firefox that fully supports seven out of the nine social plugins currently provided by Facebook, including the Like button, and partially due to API restrictions the other two. As privacy-preserving social plugins maintain the functionality of existing social plugins, we envisage that they could be adopted by social networking services themselves for the benefit of their members. To that end, we also present a pure JavaScript design that can be offered transparently as a service without the need to install any browser add-ons. Prof. Evangelos Markatos received his diploma in Computer Engineering from the University of Patras in 1988, and the M.S and Ph.D. degrees in Computer Science from the University of Rochester, NY in 1990 and 1993 respectively. Since 1992, he collaborates with the Institute of Computer Science of the Foundation for Research and Technology - Hellas (ICS-FORTH) where he is currently the founder and head of the Distributed Computing Systems Laboratory. He conducts research in several areas including distributed and parallel systems, the World-Wide Web, Internet Systems and Technologies, as well as Computer and Communication Systems Security. He has been the project manager of the LOBSTER and NoAH projects, both funded in part by the European Union and focusing on developing novel approaches to network monitoring and network security. He is currently the project manager of the i-code and SysSec projects. Since 1992, he has also been affiliated with the Computer Science Department of the University of Crete, where he is currently a full Professor. This is a Department of Computer Science Colloquium. Refreshments will be served at 11:00am Wednesday 25th April 2012 5:00pm, Ungar 411 CSC410/CSC411 Students Department of Computer Science, University of Miami will present their projects ... Olivia Lam Game Design for Metaheuristic Design Linwei Xu SAMHT Monday 23rd April 2012 5:00pm, Ungar 411 CSC410/CSC411 Students Department of Computer Science, University of Miami will present their projects ... Joel Botner 4Tap John Mauldin Independent Game Design: The Ups and Downs to Creative Freedom Joseph Chan Work Study at Computer Support of Arts and Sciences Enid Magari Internship with Claughton Capital, LLC This is another in the Department of Computer Science Pizza Seminar Series Wednesday 18th April 2012, 5:00pm, Ungar Building, Room 411 Ugan Yasavur, Reza Amini, Christine Lisetti School of Computing and Information Sciences Florida International University will present Building an On-demand Avatar-based Health Intervention for Behavior Change We discuss the design and implementation of the prototype of an avatar-based health system aimed at providing people access to an effective behavior change intervention which can help them to find and cultivate intrinsic motivation to change unhealthy lifestyles. An empathic Embodied Conversational Agent (ECA) delivers the intervention. The health dialog is informed by a computational model of Motivational Interviewing, a novel effective face-to-face patient-centered counseling style which respects an individual's pace toward behavior change. Although conducted on a small sample size, results of a user study to assess users' acceptance of the virtual counselor indicate that the system prototype is well accepted by 75% of users. This is another in the Departmental Pizza Seminar series. Thursday, April 12th, 2012, 5:00 PM, Ungar Building, Room 311 Yajie Hu Department of Computer Science University of MIami will present A Music Recommendation System Based on User Behaviors and High-level Features This thesis presents a new approach to recommending suitable tracks from a collection of songs to the user. The goal of the system is to recommend songs that are favored by the user, are fresh to the user's ear, and fit the user's listening pattern. The "Forgetting Curve" is used to assess the freshness of a song and evaluate "favoredness" using the user log. I analyze users' listening pattern to estimate the level of interest of the user in the next song. Also, user behavior is treated on the song being played as feedback to adjust the recommendation strategy for the next one. High-level features of music, like songs genre, also play an important role in music recommendation. Hence, this thesis proposes a method to classify songs in the Million Song Dataset according to song genre. Since songs have several data types, several sub-classifiers are trained by different types of data. These sub-classifiers are combined using both classifier authority and classification confidence for a particular instance. In the experiments, the combined classifier surpasses all of these sub-classifiers and the SVM classifier using concatenated vectors from all data types. Finally, I develop an application to evaluate our approach in the real world. The user logs of trial volunteers show good performance of the proposed method. This is a Department of Computer Science Masters Defense. Wednesday 28th March 2012, 5:00pm, Ungar Building, Room 411 Saman Aliari Zonouz Department of Electrical and Computer Engineering University of Miami will present Computer Security Protection via Automated and Online Intrusion Response and Recovery The severity and number of intrusions and attacks against smartphone devices as well as computer systems and networks are rapidly increasing. Preserving the availability and integrity of these computing assets in the face of fast-spreading intrusions requires advances not only in detection algorithms, but also in automated response techniques. Additionally, the rapid size and complexity growth of software systems signify the quest for systems that detect their own compromises and failures and automatically repair themselves. In particular, the ultimate goal of the intrusion response system design is to adaptively react against malicious attacks in real-time, given offline knowledge about the target system configuration, and online alerts and measurements from system-level sensors. In this presentation, we propose a new approach to automated response called the Response and Recovery Engine (RRE). Our engine employs a game-theoretic response strategy against adversaries modeled as opponents in a two-player Stackelberg stochastic game. RRE applies attack-response trees to analyze undesired security events and their countermeasures using Boolean logic to combine lower-level attack consequences. In addition, RRE accounts for uncertainties in intrusion detection alert notifications. RRE then chooses optimal response actions by solving a partially observable competitive Markov decision process that is automatically derived from attack-response trees. This is another in the Departmental Pizza Seminar series. Thursday 22nd March 2012, 5:00pm, Ungar Building, Room 402 Dr. Richard Lipton Professor and Frederick G. Storey Chair in Computing College of Computing, Georgia Institute of Technology will present The Fallacies of Interdisciplinary Research I do believe in and am excited by interdisciplinary research. But, I believe that there are many issues about such research that are misunderstood. The dictionary defines it as: "involving two or more academic, scientific, or artistic disciplines". That is true, but there is much more to it than just having two areas interact. In this talk I would like to explain my background and my past research that was interdisciplinary. I would like to highlight some of, in my opinion, the major issues that arise whenever we try to do this type of work. I will also discuss the type of teaching that is needed to make this type of research successful. Finally, I will use my crystal ball and try and predict some future interdisciplinary research directions. Predictions, especially of the future as pointed out by Niels Bohr, are always difficult and I hope that you will find my comments interesting. Dr. Richard Lipton is Stoney Chair of Computer Science at Georgia Institute of Technology. At Georgia Institute of Technology he served as Associate Dean for College of Computing from 2003 to 2008. Prior to joining Georgia Institute of Technology he was a Professor of Computer Science (now Emeritus) and a Professor of Electrical Engineering Computer Science at Princeton University, an Associate Professor of Computer Science at University of California at Berkeley, and an Associate Professor of Computer Science at Yale University. He has been a Consulting Chief Scientist of Telcordia Technologies since 1996 and has held a visiting faculty position at IBM T. J. Watson Research Center and of Xerox Parc. Dr. Lipton has made fundamental contributions to and written numerous papers in a broad range of areas in computing, including: cryptography, quantum computing, biological computation, complexity theory, network algorithms, and programming languages. Dr. Lipton received a B.S. from Case Western Reserve University in 1968 and his Ph.D. from Carnegie-Mellon University in 1973. Dr. Lipton is a Fellow of Association for Computing Machinery, a Guggenheim Fellow, and a Member of National Academy of Engineering. Wednesday 4th April 2012, 5:00pm, Ungar Building, Room 411 Javier Del Rio Manager, Operations Support*, Trade Support & Services Royal Caribbean Cruise Lines will present The Business Programmer - Mining, Analysis, and Dashboards Executives need data - constantly. They need it live, they need it analyzed, and they need it on demand. They're turning to teams of analysts to help run their business and those analysts are turning to programmers to get their tools out there 24/7. We'll take a look at technologies helping us do this, required skillsets, and how an always-on business requires always-on data. This is another in the Departmental Pizza Seminar series. Wednesday 22nd February 2012, 5:00pm, Ungar Building, Room 411 Cameron Carpenter1 and Leticia Osterberg2 and Geoff Sutcliffe1 1Department of Computer Science, 2Department of Psychology, University of Miami will present SAMHT - Suicidal Avatars for Mental Health Training Psychosocial treatments and assessments are effective for a range of psychological problems. One particular area of concern is youth suicide. This paper reports on the SAMHT intelligent tutoring system, which provides youth suicide risk assessment training. SAMHT's interactive avatar interface is based on an intelligent backend, and provides a believable interaction that is effective for training mental health professionals. This is another in the Departmental Pizza Seminar series. Wednesday 7th March 2012, 5:00pm, Ungar Building, Room 411 Shirley Cunningham, Carlos Gandarillas, and Darren North FPL Energy Services, NextEra Energy, Inc. will present Careers at NextEra Energy/FPL This is another in the Departmental Pizza Seminar series. Wednesday 15th February 2012, 5:00pm, Ungar Building, Room 411 Alfredo Weitzenfeld, PhD Professor and Director, Robotics Laboratory Division of Information Technology, College of Technology and Innovation University of South Florida Polytechnic will present Spatial Cognition in Robots based on Rat Studies The study of behavioral and neurophysiological mechanisms involved in rat spatial cognition provides a basis for the development of computational models and robotic experimentation of goal-oriented task behaviors. These models and robotics architectures offer neurobiologists and neuroethologists alternative platforms to study, analyze and evaluate new hypotheses on spatial cognition. In this talk we discuss a comparative analysis of spatial cognition in rats and robots by contrasting similar goal-oriented tasks in a cyclical maze, where studies in rat spatial cognition are used to develop computational system-level models of hippocampus integrating kinesthetic and visual information to produce a cognitive map of the environment that drives robot experimentation. During training, Hebbian learning and reinforcement learning, in the form of Actor-Critic architecture, enables a robot to learn the optimal route to a goal from a designated fixed location in the maze. During testing, robot exploits maximum expectations of reward stored within a previously acquired cognitive map to reach the goal from different starting positions. Current and future directions of this research are discussed. This is another in the Departmental Pizza Seminar series. Monday 13th February 2012, 5:00pm, Ungar Building, Room 402 Dr. George M. Ferguson Department of Computer Science, University of Rochester will present Towards Intelligent Assistants I will describe recent and ongoing research to develop intelligent software assistants that help people solve problems. Key elements of such systems include: natural, intuitive interfaces, including natural language understanding; user-centered decision support methods that support human problem-solving; and techniques to make such systems adaptable and extensible. The goal of the research is to understand the principles of such systems and how they can be applied to solve specific problems. I will demonstrate several prototypes and applications that we have developed over the years. Dr. George M. Ferguson is a research scientist in the Department of Computer Scientist at the University of Rochester. He received a B.Sc. degree in Math and Computer Science from McGill University in 1987, M.S. degrees in Computer Science from University of Alberta and from University of Rochester in 1989 and 1991, respectively, and Ph.D. in Computer Science from University of Rochester in 1995. He has served as chair, as organizer, or as a program committee member for many conferences, including Program Chair for AAAI-2004. He has proposed, received, and managed multi-million dollar research projects from NIH, DARPA, ONR, and NSF. Dr. Ferguson conducts research to bring software to the next level and to make a real difference in people's lives. This is another in the Departmental Pizza Seminar series. Wednesday 1st February 2012, 5:00pm, Ungar Building, Room 411 Steven Trac Epic, Research and Development will present The Challenges of Working in the Health IT World This talk with be an introduction to the world of health information technology. From fast retrieval and storage of patient data to government regulations and patient safety issues, the challenges never stop. I will start by explaining what is Epic, and how the HIT world is structured. Then I will describe some of the problems and challenges that we face in the health care industry. After tackling such issues, I will touch on some of the cool things we plan to do in the future. This is another in the Departmental Pizza Seminar series. Wednesday 30th November 2011, 5:00 PM, Ungar Building, Room 402 Daniel Enekes CEO of XAPT Corporation will present The Importance of the Agile ERP (Enterprise Resource Planning) Application in the Life of a Modern Business XAPT Corporation was the winner of the Microsoft Dynamics ERP Global Partner of the Year 2011 Award This is a CSMS Project Industrial Liaison Seminar. Refreshments will be served. Wednesday 30th November 2011 4:00pm, Ungar 411 CSC410/CSC411 Students Department of Computer Science, University of Miami will present their projects ... John Maulding Indie Game Development: the Modern Day David vs. Goliath At a time when the cheapest AAA video game takes roughly two years, 20 million dollars, and 70 people to make, what sort of challenges face an independent development team with limited time, money, and man power, let alone in a rapidly growing market that demands increasingly higher quality products? In this presentation, the intimidating expectations, pie eyed goals, and crushing realities facing those trying to stand toe to toe with the video game industry behemoths will be highlighted while showcasing a student team's progress in their yearlong effort to develop a video game for the Xbox 360. Sana Khan SiRNA Design and miRNA Embedding in Protein Coding Regions MicroRNAs and siRNAs ate small regulatory RNAs that play a vital role in diverse biological processes. Their ultimate function is to facilitate the reduction of the product of a target messenger RNA. Recent advances in genome synthesis technology allow for manipulation of existing genes and the creation of novel genomic sequences to specification, with a great number of applications. Incorporation of miRNA sites in mRNA sequences allows for enhanced expression control and facilitates dynamic regulatory network interactions. We explore the possibility of embedding miRNA binding sites in the coding region of a protein. This is Department of Computer Science Seminar Wednesday 2nd November 2011, 5:00pm, Ungar Building, Room 402 Dr. Briana Wellman University of Alabama will present Cooperation Paradigms for Overcoming Communication Limitations in Multirobot Wide Area Coverage Multirobot systems are an important research topic in wide area coverage applications such as hazardous waste clean-up, bomb detection, surveillance, and search and rescue missions. They can work in parallel and complete tasks faster than a single robot. Communications can support cooperation to speed up execution, reduce duplication, and prevent interference. Communication among team members is achieved explicitly or implicitly. In explicit communication, messages are intentionally transmitted and received from robot to robot. In implicit communication, robots observe the environment and other robot actions. Although many systems use explicit communications, in exploration of large, open areas (e.g. stadiums and parks), persistent intra-team digital communication is not guaranteed. Therefore, alternative approaches that do not rely upon message passing throughout exploration are needed. Novel contributions of overcoming communication limitations in wide area coverage include: (1) insight on how information shared between robots that are close has more influence on immediate action selection than information shared between robots that are farther apart. Spatial and temporal locality can be instrumental in determining relevance in subsequent action selection; (2) an approach in which observation leverages spatial and temporal locality to infer state rather than rely on digital messaging; and (3) an approach in which robots use spatial rendezvous to exchange information instead of continuously passing messages. Robots explore an environment in sectors, or designated areas, and periodically meet to communicate map information of what they explored. Simulations and physical experiments were conducted and results suggest both approaches can serve as alternatives to cooperation based on continuous point-to-point communications. Wednesday 26th October 2011, 5:00pm, Ungar Building, Room 411 Adam Pease Rearden Commerce will present Formal Ontology and the Suggested Upper Merged Ontology This talk presents an overview of ontology, including how formal ontology compares to less formal approaches and how the Suggested Upper Merged Ontology (SUMO) (www.ontologyportal.org) compares to other formal ontologies. Classes of ontology-based applications are introduced. Issues of the capabilities and tradeoffs in first order logic inference are explored. The SUMO is also described in detail, along with its mappings to the WordNet lexicon. Adam Pease is a senior researcher at Rearden Commerce in Foster City, Ca. He has led research in ontology, linguistics, and formal inference, including development of the Suggested Upper Merged Ontology (SUMO), the Controlled English to Logic Translation (CELT) system, and the Sigma knowledge engineering environment. Sharing research under open licenses, in order to achieve the widest possible dissemination and technology transfer, has been a core element of his research program and his products have been downloaded by thousands of people around the world. He is the author of the new book "Ontology: A Practical Guide". This is another in the Departmental Pizza Seminar series. Friday 21st October 2011, 10:00am, Ungar Building, Room 311 Ladislav Kavan Disney Interactive Studios will present Clothing, Skinning, and Simulating 3D Virtual Characters In this talk I will present my research on real-time 3D graphics and virtual characters. In the first part, I will summarize the motivation behind my work and my contributions to skinning, deformable collision detection, and pre-computation based rendering. In the second part, I will talk in detail about animating cloth and clothing, starting with a playback-only baking technique and subsequently moving to a more general upsampling method, which adds fine-scale details to coarse mesh simulations. Interesting technical nuggets needed to make these approaches work efficiently include: an approximate version of principal component analysis, harmonic regularization (an extension of the classical Tikhonov approach) and our take on "tracking," i.e., how to constrain fine-scale physics to follow a given coarse-scale animation. In the final part, I will conclude with some ideas for future work, revolving mostly around the question of how to make digital content creation more intuitive. Wednesday 19th October 2011, 5:00pm, Ungar Building, Room 411 Spyros Prantalos Security Software Architect, Ultimate Software will present Applied Claims-based Identity All traditional approaches to adding identity and access management functionality to applications present the same issues: they require the developers to take matters into their own hands, demand specialized security knowledge or heavily rely on the features of the underlying infrastructure. This situation has led to a proliferation of APIs and techniques, forcing developers to continually re-learn how to perform the same task with different APIs or in other words constantly "reinventing the wheel". Claims-based identity is an approach that changes the way we think about authentication and authorization, adding a logical representation of identity transactions and identify the roles that every entity plays. By adding that further level of indirection, claims-based identity created the basis for decoupling the programming model and the details of the deploy-time systems. Consequently, taking the claims-based identity approach we can achieve high-level yet accurate descriptions, irrespective of their implementation details. Note: This will be an excellent venue to learn about a local software company that hires interns and full time software engineers. Undergraduate Computer Science students are especially encouraged to attend. This is another in the Departmental Pizza Seminar series. Wednesday 12th October 2011, 5:00pm, Ungar Building, Room 402 Ubbo Visser Department of Computer Science, University of Miami will present Knowledge Representation and Reasoning for Multiagent Systems and Semantic Web Applications The area of Artificial Intelligence is diverse and at a crossroads. Much of the past research in the area has focused either on high-level reasoning from abstract, ungrounded representations or on interpreting raw sensor data towards building grounded representations. However, neither of these foci taken alone is sufficient for deploying practical, real-world AI systems. My research focuses on knowledge representation and reasoning in two application areas: (1) Agents in Dynamic and Real-Time Environments and the (2) Semantic Web. I will address both areas: (1): In recent years, much work has been done on creating complete autonomous agents: those that sense their environment, engage in high-level cognitive decision-making, and then execute their actions in the environment. We will explore the problem of enabling autonomous agents in finding the right position on the field or in tracking the ball within a soccer game while dealing with time constrains, adverse opponents, and dynamic environments. This research has implications for a variety of autonomous agents and robotics, such as robots that assists humans on an every day basis. (2): The focus of my research in the Semantic Web area is the development of methods in order to solve the semantic heterogeneity problem (e.g., catalogue integration). I concentrate and evaluate on emerging ontology-based approaches in order to find and integrate information sources for the construction of ontologies. These techniques are used to solve problems that can be described with terminological logics. I will demonstrate an example from our NIH funded research project "BioAssay Ontologies" (in collaboration with the Miller School of Medicine). Wednesday 5th October 2011, 5:00pm, Ungar Building, Room 411 Geoff Sutcliffe Department of Computer Science, University of Miami will present Combining Proofs to form Different Proofs Different Automated Theorem Proving (ATP) systems solve different parts of different problems in different ways. Given a set of proofs produced by ATP systems based on adequately common principles, it is possible to create new proofs by combining proof components extracted from the proofs in the set. It is not generally easy to say that one of the original or new proofs is better or worse than another, but ways to show that two proofs are different are available. This talk describes a process of proof combination to form new proofs that are different from the original set of proofs. This is another in the Departmental Pizza Seminar series. Friday 16th September 2011, 3:00pm, Ungar Building, Room 411 Stephan Schürer, Ubbo Visser, Saminda Abeyruwan, Dusica Vidović, Barun Bhhararai, Uma Vempati Department Molecular and Cellular Pharmacology, Department of Computer Science, Center for Computational Science University of Miami will present Development and Application of Semantic Web Technologies to Biomedical Research and Chemical Biology Modern life science and drug discovery research generates very large amounts of so called -omics data, including data about properties and functions of genes, proteins, and more complex biological entity modules. Large amounts of data are also generated of perturbations of biological model systems for example by chemicals. All with the goal to better understand the mechanisms of human disease and how dysregulated systems can be put back into normal function. However, despite the exponential increase of available data, new knowledge does not appear to be generated at the same pace, as can be seen for example at the relatively constant rate of new approved drugs over the last 50 years. The enormous complexity of biological systems and mechanisms of disease call for novel approaches combining various experimental and computational strategies and require interdisciplinary cross training and collaboration. The University of Miami with the medical school and the College of Arts and Sciences provides an ideal environment to achieve just that. For about 18 months, via the Center for Computational Science, the research groups of Dr Visser and Dr Schürer have been collaborating to develop novel approaches to enable researchers to better utilize existing high-throughput datasets. This collaboration has been so successful that we are now planning to expand it further while providing opportunities for students who are interested to develop the cross-disciplinary expertise required to address the most pressing issues related to human health. In this seminar we will briefly introduce several current research projects covering different aspects of drug discovery. A unifying aspect is the goal of enabling scientists to better utilize large and complex available datasets and the development and application of semantic web technologies to achieve this goal. We will introduce 1) the BioAssay Ontlogy (BAO) project (http://bioassayontology.org/), which is focused on categorizing and analyzing biological screening data; 2) "Phylostructeomics" - proteome-wide analysis of binding sites and ligand protein binding interactions targeted toward poly-pharmacology; and 3) the SMARTNames project focused on functional chemistry space, chemical reactivity and synthetic chemistry. More information and publications can be found at http://biomed.miami.edu/sschurer and http://www.ccs.miami.edu/chemoinformatics-team.html. This is another in the Departmental Pizza Seminar series. Wednesday 7th September 2011, 5:00pm, Ungar Building, Room 411 Dr. Carlos G. Gallo Center of Computational Science University of Miami will present An Information Theoretical View of Dialogue Psycholinguistic research shows wide evidence of the online, incremental nature of language understanding. Artificial systems that interpret language, however, have generally been designed based on essentially offline architectures. This talk is divided into two parts. First, we present a corpus of task-oriented interactive dialogue that is used to develop conversational agent systems. Second, we use information theory to analyze to what extend speakers follow optimal strategies in distributing information during language production. This is another in the Departmental Pizza Seminar series. Friday, July 15th 2011, 3:00pm, Ungar Room 402 Rahul Chawla Department of Computer Science University of Miami will present Filtering Social Tags for Songs based on Lyrics using Clustering Methods In the field of Music Data Mining, Mood and Topic information has been considered as a high level metadata. The extraction of mood and topic information is difficult but is regarded as very valuable. The immense growth of Web 2.0 resulted in Social Tags being a direct interaction with users (humans) and their feedback through tags can help in classification and retrieval of music. One of the major shortcomings of the approaches that have been employed so far is the improper filtering of social tags. This thesis delves into the topic of information extraction from songs' tags and lyrics. The main focus is on removing all erroneous and unwanted tags with help of other features. The hierarchical clustering method is applied to create clusters of tags. The clusters are based on semantic information any given pair of tags share. The lyrics features are utilized by employing CLOPE clustering method to form lyrics clusters, and Naive Bayes method to computer probability values that aid in classification process. The outputs from classification are finally used to estimate the accuracy of a tag belonging to the song. The results obtained from the experiments all point towards the success of the method proposed and can be utilized by other research projects in the similar field. This is a Department of Computer Science Masters Defence. Friday, May 6th, 2011, 2:00 PM, Ungar Building, Room 402 Dr. M. Brian Blake College of Engineering University of Notre Dame will present Service-Oriented Computing: Emerging Approaches for Web-Based Software Engineering Emerging technologies facilitate an environment where web-based software or web services have well-defined, open interfaces and are discoverable across the Internet. Service-oriented computing is an emerging approach to software engineering that suggests that new specialized business processes can be created, on-demand, simply by integrating the services provided by others. However, in the real world, software developers tend to create applications that do not conform to consistent developmental practices even if they do use universal interface representations (e.g. the eXtensible Markup Language). Our research utilizes semantic approaches, enhanced syntactical methods, and contextual information to automate the integration of software services that are developed randomly from a wide array of diverse sources. This talk discusses our foundational lines of research and subsequent contributions in the areas of service discovery, composition, and evaluation. The talk will conclude with future work that leverages service-oriented paradigms in areas such as visual analytics, smart grid, and the "Internet of things". This is a Department of Computer Science Colloquium. Wednesday 20th April 2011, 5:00pm, Ungar Building, Room 411 Benoit Razet will present Computing Certificates of Regular Expressions Equivalence Regular expressions are mostly known as pattern matching expressions in scripting languages (Perl, sed, awk, etc.). They are also theoretically studied as being part of Automata Theory. A regular expression denotes a language (set of words) and two regular expressions are equivalent when they denote the same language. The equivalence problem is decidable and furthermore, any equivalence can be proved axiomatically. Various such axiomatisations have flourished during the last fifty years, we will survey the most important ones. Every axiomatisation comes together with a proof of completeness of the following form: if two regular expressions are equivalent then there exists a proof within the axiomatic system. In principle, when this completeness proof is constructive one can extract from it an algorithm that produces a certificate of the equivalence. We have filled the gap developing a program (written in OCaml) that computes certificates. For this purpose we have designed a domain specific proof system with a language of commands for composing a certificate checkable within this system. During the talk I will make a demonstration of the program. This is a joint work with Bodhayan Roy at TIFR. This is another in the Departmental Pizza Seminar series. Wednesday, April 13th, 2011, 5:00 PM, Ungar Building, Room 411 Justin Stoecker Department of Computer Science University of MIami will present TerraVis: A Stereoscopic Viewer for Interactive Seismic Data Visualization Accurate earthquake prediction is a difficult, unsolved problem that is central to the ambitions of many geoscientists. Understanding why earthquakes occur requires a profound understanding of many interrelated processes; the Earth functions as a massive, complex system. Scientific visualization can be applied to such problems to improve understanding and reveal relationships between data. There are several challenges inherent to visualizing seismic data: working with large, high-resolution 3D and 4D data sets in a myriad of formats, integrating and rendering multiple models in the same space, and the need for real-time interactivity and intuitive interfaces. This work describes a product of the collaboration between computer science and geophysics. TerraVis is a real-time system that incorporates advanced visualization techniques for seismic data. The software can process and efficiently render digital elevation models, earthquake catalogs, fault slip distributions, moment tensor solutions, and scalar fields in the same space. In addition, the software takes advantage of stereoscopic viewing and head tracking for immersion and improved depth perception. During reconstruction efforts after the devastating 2010 earthquake in Haiti, TerraVis was demonstrated as a tool for assessing the risk of future earthquakes. This is a Department of Computer Science Masters Defense. Tuesday 12th April 2011, 5:00 PM, Ungar Building, Room 402 James Vallord-Costa CEO of Hypernia Hosting will present How to Make your First Million on the Internet James shares how he went from making pennies as a local Miami web developer to building the largest video game hosting company in the world in less than 8 years. James is the CEO of Hypernia, and specializes in hosting the world's top media and entertainment businesses. He is responsible for connecting over 50 million people together each and every day through social networks, online games, websites and online communications. His goal is to keep technology affordable and flexible for everyone to promote social progress. In 2008, he worked with Michael Moore in distributing "Slacker Uprising" online, breaking all worldwide download records for a feature film. This is a CSMS Project Industrial Liaison Seminar. Refreshments will be served. Monday, April 4th, 2011, 5:00 PM, Ungar Building, Room 345 Sander van Dijk and the RoboCanes team Department of Computer Science University of Miami will present RoboCup Soccer Simulation 3D: Competition and Collaboration The RoboCup Soccer Simulation 3D teams RoboCanes, of the University of Miami, and Bold Hearts, of the University of Hertfordshire, have joined forces under a grant of the RoboCup federation to enhance and develop the infrastructure for one of the largest robotics and AI competitions in the world. Firstly, we made the simulation system used in the simulated humanoid robotic soccer competitions ready to enable competitions on a larger scale and more dynamic than the years before. Secondly, we have designed and developed a framework to perform massive parallel learning and optimisation in the complex scenario of robotic soccer. Despite this collaboration, the teams are still competitors, currently competing for the first prize in the German Open competitions. In this seminar we will give an overview and demos of both our collaborative and competitive work. This is a Department of Computer Science Colloquium. Refreshments will be served. Friday, April 1st, 2011, 5:00 PM, Ungar Building, Room 402 Mark Crovella Professor of Computer Science Boston University will present Inferring Invisible Internet Traffic The Internet is at once an immense engineering artifact, a pervasive social force, and a fascinating object of study. Unfortunately many natural questions about the Internet resist precise answers, for a variety of reasons. In this talk I'll start by discussing why measuring the Internet's infrastructure, traffic, and applications is interesting and important, and why it is often so difficult. I'll then touch on methods that are often used to overcome the Internet's resistance to being measured -- in particular, statistical inference. As a detailed example I'll describe a current project in traffic measurement. We are asking the question: using traffic measurements taken at one location in the Internet, can we estimate how much traffic is flowing in a different part of the Internet? I'll show evidence that such estimation is indeed possible, and suggest how this technique may be used to answer both scientific and commercial questions. Mark Crovella is Professor of Computer Science at Boston University. He received his Ph.D. in Computer Science from the University of Rochester in 1994. His research interests are in performance evaluation, focusing on parallel and networked computer systems. In the networking arena, he has worked on characterizing the Internet and the World Wide Web. He has explored the presence and implications of self-similarity and heavy-tailed distributions in network traffic and Web workloads. He has also investigated the implications of Web workloads for the design of scalable and cost-effective Web servers. In addition he has made numerous contributions to Internet measurement and modeling; and he has examined the impact of network properties on the design of protocols and the construction of statistical models. Professor Crovella is co-author of "Internet Measurement: Infrastructure, Traffic, and Applications" (Wiley Press, 2006) and is the author of over one hundred papers, with over 13,000 citations (Google Scholar, 2010). Between 2007 and 2009 he was Chair of ACM SIGCOMM. In 2010 he received the SIGMETRICS Test of Time Award for "Self-Similarity in World Wide Web Traffic: Evidence and Possible Causes". Professor Crovella is a Fellow of the ACM. This is a Department of Computer Science Colloquium. Refreshments will be served. Wednesday 23rd March 2011, 5:00pm, Ungar Building, Room 411 Andreas Seekircher Department of Computer Science, University of Miami will present Entropy-based Active Vision for a Humanoid Soccer Robot Every agent needs information about the current state and the environment to act successfully. Often retrieving information from sensor measurements is considered as a passive process, that is done while the agent tries to reach a goal. However in many scenarios the actions of an agent have an influence on the perceptions. So it is possible to consider the information gain in the selection of actions, such that the error in the estimated state is decreased. Strategies for active sensing or active localization can lead to a large improvement of the overall performance. The RoboCup is a robotics competition where autonomous robots play soccer. A soccer robot in this environment can not perform well without information about the own position and the location of the ball. Smaller errors in the state estimation lead to a more accurate behavior. A simple head movement that scans the environment in a fixed pattern already works, but it has to be tuned manually and is not flexible. It will be shown how the estimation of a robot's world model can be improved by actively sensing the environment through considering the current world state estimate through minimizing the entropy of an underlying particle distribution. Being originally computationally expensive, this approach was optimized to become executable in real-time on a robot with limited resources. This approach was used on a humanoid robot, performing self-localization and ball tracking on a RoboCup soccer field. Andreas Seekircher is a PhD student at the University of Miami. He holds a M.Sc. in Computer Science from the University of Bremen, Germany. The paper "Entropy-based Active Vision for a Humanoid Soccer Robot" won the best paper award on the RoboCup Symposium 2010. His research is focused on robotics and artificial intelligence, specifically controlling autonomous robots. This is another in the Departmental Pizza Seminar series. Wednesday 9th March 2011, 5:00pm, Ungar Building, Room 411 Sander van Dijk University of Hertfordshire will present Grounding Behavior in Informational Bounds Acquiring and processing information is important for agents, both natural as artificial. Being able to use more information can give the agent an advantage over others. However, information can be very costly for many organisms, from blow flies to humans, it is known that information related processes can take up over 20% of the full energy consumption. So, we can assume that an agent tries to maximize its informational capability, but only up to the minimal requirements for being able to act successfully. We can study the effect of this assumption, and of informational bounds in general, on behavior, using tools from the well-established field of Information Theory. I will discuss some of the aspects and behavioral properties arising from this methodology, such as Empowerment, which measures the amount of observable control of an agent on the environment and leads to intrinsically motivated behavior, and Relevant Information, which is the minimum amount of information needed to achieve a certain level of performance on a task. Specifically, I will focus on hierarchical behavior structures, which are popular in ethology and machine learning, although deep understanding of why they are attractive, why they work, and whether they are necessary, is missing. Starting from the argument that behaviour generation is constrained by a range of limits on information acquisition and processing, I show how such limits spontaneously give rise to abstraction and structuring of tasks and behaviour generation, and thereby suggest that hierarchical structuring is a natural way of coping with informational constraints. This is another in the Departmental Pizza Seminar series. Wednesday 2nd March 2011, 5:00pm, Ungar Building, Room 411 Blanca Abraham Computer Science Department University of Los Andes will present Intelligent Architecture for Software Development Monolithic programs are now being gradually replaced with software that is made up of a number of components, originating from various sources. Every software component is associated not only with functional, but also with non-functional aspects. A reflective middleware framework has been built to capture characteristics that a software component should have and how other components depend on those characteristics to keep the whole system working as required. The main idea is to monitor (introspection) the applications, in order to change or tune (intercession) the software components with the specific goal of making the whole system workable. To find new components, two stochastic search algorithms were developed and added to the middleware. Software component selection is done using both the most relevant characteristics that the component should have so they can be integrated to the other existing components and swarm intelligence techniques; specifically pheromone value. This is another in the Departmental Pizza Seminar series. Wednesday 23rd February 2011, 5:00 PM, Ungar Building, Room 506 Thomas Duxbury Physics and Astronomy Department George Mason University will present Computer Science in Planetary Exploration Computer Science dominates every aspect of robotic and human exploration of the planets. Prior to launch, mission design and planning merge planetary body models with predicted launch vehicle, orbit, spacecraft and science payload instrument models to optimize the probability of success, define mission phases and science observation sequences, and minimize risks by having sufficient weight, power, data transmission, etc. resource margins to handle the "unknown" unknowns. During flight operations, these same models and algorithms are updated to reflect the knowledge learned of the planetary bodies, instrument flight performances and anomalies encountered to continually re-plan the remainder of the mission. Finally the scientific data analyses is performed to glean every tidbit of information that will help us determine the origins, evolutions and current states of the planetary bodies in our solar system and what lies ahead for humans. Examples will be given on observation planning, instrument modeling, image restoration, diverse data set registration and map projection, data visualization and flight operations. At the heart of all of these are models, data and algorithms optimized to run on computers, with parallel computers becoming more prevalent, having print, plot and interactive voice and terminal displays (i.e., computer science) to learn the secrets hidden in the planetary bodies. This is a CSMS Project Industrial Liaison Seminar. Refreshments will be served. Thursday 17th February 2011, 5:00 PM, Ungar Building, Room 402 Facundo Memoli Mathematics Department Stanford University will present Metric Structures on Datasets: Design, Stability, and Classification of Algorithms for Data and Shape Analysis Several methods in data and shape analysis can be regarded as transformations between metric spaces and in general, as maps from k-tuples of metric spaces into a metric space. Examples are hierarchical clustering methods, the higher order constructions of computational persistent topology, and several computational techniques that operate within the context of data/shape matching under invariances. Metric geometry, and in particular different variants of the Gromov-Hausdorff distance provide a point of view which is applicable in different scenarios. The underlying idea is to regard datasets as metric spaces, or metric measure spaces (a.k.a. mm-spaces, which are metric spaces enriched with probability measures), and then, crucially, at the same time regard the collection of all datasets as a metric space in itself. Variations of this point of view give rise to different taxonomies that include several methods, both preexisting and novel, for extracting information from datasets and shapes. I will show an ongoing application of one of these new methods to functional classification of chemical compounds by shape properties. Imposing metric structures on the collection of all datasets could be regarded as a "soft" construction. The classification of certain algorithms, or the axiomatic characterization of them, can be achieved by imposing more "rigid" directed-graph structures on the collection of all finite metric spaces. I will describe some results along these lines in the context of clustering schemes. This is a Department of Computer Science Colloquium. Refreshments will be served in UB345 at 4:30pm. Monday 14th February 2011, 5:00 PM, Ungar Building, Room 311 Miriah Meyer Postdoctoral Research Fellow Harvard University will present Visualizing Biological Data Visualization tools are essential for deriving meaning from the avalanche of data we are generating today. To facilitate an understanding of the complex relationships embedded in this data, visualization research leverages the power of the human perceptual and cognitive systems, encoding meaning through images and enabling exploration through human-computer interactions. In my research I design visualization systems that support exploratory, complex data analysis tasks by biologists who are analyzing large amounts of heterogeneous data. These systems allow users to validate their computational models, to understand their underlying data in detail, and to develop new hypotheses and insights. My research process includes five distinct stages, from targeting a specific group of domain experts and their scientific goals through validating the efficacy of the visualization system. In this talk I'll describe a user-centered, methodological approach to designing and developing visualization tools and present several successful visualization projects in the areas of genomics and systems biology. I will also discuss the long term implications of this work and the field of visualization as a whole. This is a Department of Computer Science Colloquium. Refreshments will be served in UB345 at 4:30pm. Wednesday 9th February 2011, 5:00pm, Ungar Building, Room 411 Nate Blaylock Florida Institute for Human and Machine Cognition (IHMC) will present Geolocating Text and Extracting Medical Histories Although there is a large amount of information available in formats suitable for machine reasoning (such as relational databases), the vast majority of human-made information is only available as natural language text. In this talk, I will present a hybrid, deep natural language processing approach to Information Extraction and discuss its application in two domains. First, I will describe a system which extracts geospatial information mentioned in text. Using a combination of the text mention as well as a geographic information system (GIS) database, the system grounds this information in geospatial entities and latitude and longitude coordinates and can track an entity's location over time, based only on the text description. I will also describe our recent preliminary work on using this same deep NLP framework for extracting medical history information from clinical texts. Nate Blaylock received a Ph.D. in Computer Science at the University of Rochester in 2005 where his dissertation dealt with natural language understanding and plan recognition for dialog systems. Prior to joining IHMC in 2007, Nate was a Research Scientist at Cycorp, where his research focused on using predictive event models to apply plan recognition to level 2/3 fusion problems. Prior to that, Nate spent two years in a postdoctoral position at Saarland University in Saarbrucken, Germany, where he helped build the bilingual German/English SAMMIE in-car dialog system, using the theory of collaborative problem solving from his thesis to understand and reason with natural language in dialog. At IHMC, Nate continues to do research on dialog systems, plan and intent recognition, and natural language understanding, including information extraction in the medical and geospatial domains. This is another in the Departmental Pizza Seminar series. Thursday 3rd February 2011, 5:00 PM, Ungar Building, Room 402 Mark Crovella Professor of Computer Science Boston University will present Inferring Invisible Internet Traffic The Internet is at once an immense engineering artifact, a pervasive social force, and a fascinating object of study. Unfortunately many natural questions about the Internet resist precise answers, for a variety of reasons. In this talk I'll start by discussing why measuring the Internet's infrastructure, traffic, and applications is interesting and important, and why it is often so difficult. I'll then touch on methods that are often used to overcome the Internet's resistance to being measured - in particular, statistical inference. As a detailed example I'll describe a current project in traffic measurement. We are asking the question: using traffic measurements taken at one location in the Internet, can we estimate how much traffic is flowing in a different part of the Internet? I'll show evidence that such estimation is indeed possible, and suggest how this technique may be used to answer both scientific and commercial questions. Speaker Bio: Mark Crovella is Professor of Computer Science at Boston University. He received his Ph.D. in Computer Science from the University of Rochester in 1994. His research interests are in performance evaluation, focusing on parallel and networked computer systems. In the networking arena, he has worked on characterizing the Internet and the World Wide Web. He has explored the presence and implications of self-similarity and heavy-tailed distributions in network traffic and Web workloads. He has also investigated the implications of Web workloads for the design of scalable and cost-effective Web servers. In addition he has made numerous contributions to Internet measurement and modeling; and he has examined the impact of network properties on the design of protocols and the construction of statistical models. Professor Crovella is co-author of "Internet Measurement: Infrastructure, Traffic, and Applications" (Wiley Press, 2006) and is the author of over one hundred papers, with over 13,000 citations (Google Scholar, 2010). Between 2007 and 2009 he was Chair of ACM SIGCOMM. In 2010 he received the SIGMETRICS Test of Time Award for "Self-Similarity in World Wide Web Traffic: Evidence and Possible Causes". Professor Crovella is a Fellow of the ACM. This is a Department of Computer Science Colloquium. Refreshments will be served. Wednesday 8th December 2010, 5:00 PM, Ungar Building, Room 402 Gerhard Kraetzschmar Professor for Autonomous Systems Bonn-Rhein-Sieg University will present Software Development in Robotics Developing software to bring a complex mobile robot system to life and have it exhibit intelligent behavior in an autonomous, robust, and safe manner is a very, very difficult, complex, time-consuming, and error-prone endeavor. In this talk I analyze the reasons and origins of this complexity and present some ideas for handling it. One topic of particular relevance is architecture, and its manifold interpretations of it. I survey the approach we take in the project BRICS to structure architectural issues. Finally, I present a software development process model we recently developed in order to organize software development in robotics such that it meets professional standards. Gerhard K. Kraetzschmar is a professor for autonomous systems at Bonn-Rhein-Sieg University, Germany. He obtained a M.Sc. (Diploma) and a Ph.D. (Dr.-Ing.) in Computer Science, both from University of Erlangen. His research interests include a wide range of fields in robotics, artificial intelligence, artificial life, and neuroinformatics, with special focus on software development for robotics, multirobot teams, robot learning, middleware for robotics, and educational robotics. He is a member of AAAI, ACM, IAS, IEEE, and GI, and Vice President RoboCupJunior of the RoboCup Federation. This is a joint Department of Computer Science and Department of Electrical and Computer Engineering Colloquium. Refreshments will be served. Wednesday 1st December 2010 4:30pm, Ungar 411 CSC410/CSC411 Students Department of Computer Science, University of Miami will present their projects ... Adam Schwartz - Facial Motion Capture for Games Pablo Bolivar Morales - Estimation of Repeats from Sparce Next Generation Sequencing Andres Acevedo Ulloque - Android Game Development: The Runner This project explores development for mobile platforms and the trend that follows it. We will talk about the use a game engine (AndEngine) during the development of a game for the Android platform. Additionally, we will discuss from the design concept to the implementation, the difficulties and challenges faced during the process of developing an Android game. Demo of Android game - Runner. This is another in the Departmental Pizza Seminar series. Monday 29th November 2010 4:30pm, Ungar 411 CSC410/CSC411 Students Department of Computer Science, University of Miami will present their projects ... Joel Botner - Input Abstraction: Beyond the QWERTY Keyboard Gracia Bonilla - CoSeChar: Genomic Coding Sequence Characterization Tools Cameron Carpenter - SporkleAI: Automatic Interfacing and Problem Solving on Sporkle.com Oscar Sanchez - Upgrading and Restructuring the "Save Dade" Website and Services This is another in the Departmental Pizza Seminar series. Monday 22nd November 2010, 5:00 PM, Ungar Building, Room 506 Dr. Ashwanth Srinivasan Rosenstiel School of Marine and Atmospheric Science will present Many Task Computing for Modeling the Fate of Oil Discharged from the Deep Water Horizon Well Blowout The Deep Water Horizon well blowout on April 20th 2010 discharged between 40,000 - 1.2 million tons of crude oil into the Gulf of Mexico. In order to understand the fate and impact of the discharged oil, particularly on the environmentally sensitive Florida Keys region, we have implemented a multi-component application which consists of many individual tasks that utilize a distributed set of computational and data management resources. The application includes two 3D ocean circulation models of the Gulf and South Florida and a multi-phase oil spill model. Information from ocean models is integrated with a multi-phase oil model that tracks the fate of approximately 10 million oil particles. These individual components execute as large parallel applications on a 640 core IBM Power 5 cluster and a 5040 core Linux cluster, both operated by the Center for Computational Science, University of Miami. The work flow between the models is handled by a custom distributed software framework built using the Open Project for Networked Data Access Protocol (OPeNDAP). In this talk, we present sample results and discuss computational challenges in deploying such many task distributed work flows on large clusters and heterogeneous architectures. This is a CSMS Project Industrial Liaison Seminar. Refreshments will be served. Wednesday 10th November 2010, 5:00pm, Ungar Building, Room 411 Dimitris Papamichail Department of Computer Science, University of Miami and Scott Baker Division of Marine Geology and Geophysics will present GPU Computing: Architecture, Algorithmic Techniques and Programming GPU computing is the use of a GPU (graphics processing unit) to perform general purpose scientific and engineering computations. The GPU has evolved over the years to have teraflops of floating point performance. New massively parallel architectures enable the use of hundreds of processor cores for general computations, while their modest price and accessible programming interfaces allow the general users to take advantage of these developments. In this seminar we will examine the modern GPU architecture and design philosophy, exposing some of the benefits and limitations of the platform. We will introduce algorithmic techniques for this specialized architecture and demonstrate how to create programs using the CUDA programming model. This is another in the Departmental Pizza Seminar series. Wednesday 3rd November 2010, 5:00pm, Ungar Building, Room 411 Ryan Thomas Brown Department of Computer Science, University of Miami will present Synthetic: A New Interface for Sound Design Modular audio programming environments such as SuperCollider or Max/MSP are used for a wide variety of tasks, such as algorithmic music composition, live sound and art installations, and acoustics research. They allow users to quickly develop specialized tools in an environment tailored for working with audio. While they are enormously powerful, using them for sound design is sometimes difficult because of the lack an intuitive interface for exploring the parameters of the synthesis graphs built in them. Synthetic is a program that wraps around SuperCollider and provides this interface. It aims to be a new breed of software that is a cross between a wave editor and a modular audio environment. This is another in the Departmental Pizza Seminar series. Wednesday 27th October 2010, 5:00pm, Ungar Building, Room 411 Adam McMahon Department of Computer Science, University of Miami will present Rendering Animations for Social Volunteer Computing While both volunteer computing and social networks have proved successful, the merging of these two models is a new field: Social Volunteer Computing. A Social Volunteer Computing system utilizes the relationships within a social network to determine how computational resources flow towards tasks that need to be completed, and the results of these computations are added back into the social network as content. Such a system will provide scientists and artists a new facility to obtain computational resources and disseminate their work. RenderWeb, a prototype Social Volunteer Computing system, is introduced that allows animations created in Blender to be distributed and rendered within Facebook. This is a Department of Computer Science PhD Proposal Defense Wednesday 29th September 2010, 5:00pm, Ungar Building, Room 411 Justin Stoecker Department of Computer Science, University of Miami will present Interactive Volume Rendering for Scientific Visualization The traditional rendering technique of real-time computer graphics involves the use of geometric primitives, such as triangles, to construct surface representations of objects. This technique works well in many situations, especially when drawing objects that are solid and have easily recognizable surfaces. However, there are many visual phenomena that are nebulous, contain partial transparency, or are too complex to effectively visualize with polygonal meshes. Examples found in nature include smoke, fire, cloudes, and crepuscular rays ("god rays"). Scientists and engineers also require methods for interpreting volumetric data: pressure measurements in an area surrounding a fault, air turbulence, the behavior of fluids, or medical data with several layers and materials. With the recent evolution of programmable graphics hardware, volumetric data and related phenomena can be rendered with interactive framerates. This presentation will provide a high-level overview of the programmable graphics pipeline and how it enables powerful new approaches to visualizing volume data. The seminar will focus on 3D texture-based GPU volume rendering and demonstrate a few applications within scientific visualization. This is another in the Departmental Pizza Seminar series. Wednesday 28th April 2010 4:30pm, Ungar 411 CSC410/CSC411 Students Department of Computer Science, University of Miami will present their projects ... Adam Schwartz - Motion Capture-based Facial Animation for Games. Facial animations are an important element in modern video games. Prof. Visser's soccer simulator lacks this. Facial motion capture can be used to this end, but requires a good amount of processing before it can be used in a game. It must be properly mapped, and the mappings must be translated properly into animation. While implementing this with predefined blendshapes may appear to be the simpler task, this is not the case if the animations are intended to be used with a very large number of characters. As such, a method using a bone structure, while perhaps less visually appealing, is a far more practical solution. Michael Schick - The Language of the Future. Have you ever wanted to ask Google a math problem, or the name of the Nobel Peace Prize winner from 1919, and get back the answer not just a link to a "relevant" website? Automated Theorem Proving provides us this possibility! In my talk I will be discussing the advantages of Automated Theorem Proving, specifically the advantages of certain languages, and how my research is helping to improve the field. Matthew Ferens - Working in the Unreal Engine. The concept of video games has become an integral part in any Computer Science discussion as it incorporates many aspects of the field into a multi-billion dollar industry. Video game engines, powerful platforms on which games are built, were once extremely restricted and proprietary, but many have recently been released for public consumption. The Unreal Development Kit is perhaps the best known and most powerful as it allows for creation of everything from simple levels to complete games by anyone with the will to learn. The purpose of this project was to work with the engine and gain experience that can apply to both scholarly and real-world work. The project focused on simple level design, scripting, and connecting it all together into a final product. Benjamin Drolet - Reducing Spam on Online Forms. Have you ever managed a web page with form validation? Have you received annoying SPAM from people blasting your forms? Have you ever had to fill out an annoying form online? My talk will discuss a new way that I developed to reduce SPAM through forms. I will talk about the advantages and disadvantages of form validation and how my program is helping improve the process. This is another in the Departmental Pizza Seminar series. Wednesday 21st April 2010, 5:00pm, Ungar Building, Room 402 Nelson Dellis Department of Computer Science, University of Miami will present Remembering not to Forget Do you often misplace your keys? Forget someone's name a second after you've heard it? Wish you could memorize facts for a test quicker? If the answer to any of those is yes, then you are probably in the same boat as everyone else! But it doesn't have to be that way. I will be giving a seminar on how to improve your memory drastically. Remembering names, numbers, words, can be easy and fun. I have been able to train my brain to memorize a 200 digit number in less than 5 minutes, the order of a shuffled deck of cards in just under a minute, and the names of everyone I meet. In this talk I will be discussing the techniques I have trained my memory to use in order to accomplish these amazing feats. The best part: anyone can learn them! See Nelson on TV. This is another in the Departmental Pizza Seminar series. Wednesday 14th April 2010, 5:00pm, Ungar Building, Room 411 Mortadha Alkhawaja Department of Computer Science, University of Miami will present Semantic Search in Geospatial Knowledge Bases Geospatial data integration with semantic web technology is one of the hot topics in the knowledge and data engineering field. What makes the integration of geospatial data task more complicated than others is the variety of the geospatial data models, formats, semantics and relations. Moreover, in order to employ the web as a medium for data and information integration, comprehensive datasets and vocabularies are required as they enable the disambiguation and alignment of other data and information. All these factors are considered challenges in the design and implementation of a semantic web based GIS system. TerraFly, a public service of Florida International University sponsored by NSF (MII, MRI et al.) and by NASA, USGS, and IBM is our candidate on which we apply the semantic web techniques for the purpose of integrating data from various knowledge bases, and enriching its search service with more accurate and meaningful query results. This is another in the Departmental Pizza Seminar series. Tuesday 6th April 2010, 5:00 PM, Ungar Building, Room 402 Syed Waqar Ali, Ph.D. Hurricane Engineering & Testing, Inc. will present Building Products Testing Post Hurricane Andrew In response to Hurricane Andrew, South Florida Building Code was revised to include debris impact protection of products covering exterior openings of the building. The new standards also implemented ASCE-7-88 which increased the wind pressure on building envelopes. The new testing standards with enhanced pressures, required development of advanced testing and analysis facilities. The task required the merger of several engineering and scientific disciplines. We will examine how an interdisciplinary approach to problem solving was used to address these challenges. This is a CSMS Project Industrial Liaison Seminar. Refreshments will be served. Wednesday 31st March 2010, 5:00pm, Ungar Building, Room 345 Justin Stoecker Department of Computer Science, University of Miami will present Stereo Visualization of the Haiti Earthquake The recent earthquake in Haiti has motivated scientists and engineers to assess the risk of future earthquakes that could affect the rebuilding of Port-au-Prince. As part of a collaborative effort with geophysicists in the Rosentiel School of Marine and Atmospheric Science, we have developed prototype software for visualizing seismic data. Our software takes advantage of stereoscopic 3D rendering for immersion and improved perception of depth. This talk will describe the relevance of the data being visualized as well as the details of the software's implementation. The fundamentals of stereo vision and differences between the various stereo technologies will also be covered. This seminar will involve an interactive demonstration of the stereoscopic 3D vision. This is another in the Departmental Pizza Seminar series. Wednesday March 10th 2010 5:00pm, Ungar 411 Dr. Ziya Arnavut Department of Computer & Information Sciences, SUNY Fredonia will present Permutation Codes for Image Compression Images contain large amount of data and are used in many applications. Not only compressed image data save space, but also in certain applications such as the World-Wide-Web (WWW), they save time since the amount of data to be transmitted is smaller than the original image. Codebook approaches have been used in lossy image compression. For example, vector quantization employs codebook. For lossless compression of images, more than a decade ago, a codebook approach based on spanning trees was proposed. In this talk we address suitability of permutation codes for still images. This is another in the Departmental Pizza Seminar series. Wednesday February 24th 2010 5:00pm, Ungar 411 Nelson Dellis Department of Computer Science, University of Miami will present Using Controlled Natural Language for Reasoning with World Knowledge Search Engines are our number one source for finding answers to questions in the 21st century, but unfortunately they do not provide us with immediate answers. The answers still need to be extracted by the user from the results returned by the search engine. This talk addresses this problem and discusses how we are getting closer to being able to reason about world knowledge by using theorem provers and how they provide us with the exact answers we are looking for. Currently, the theorem provers used to provide this service lack a user-friendly means of submitting a query. This talk will also discuss how, by using controlled natural language, one can submit queries with ease and in a natural way rather than a complex and esoteric way. This is another in the Departmental Pizza Seminar series. Wednesday February 3rd 2010 5:00pm, Ungar 411 Adam McMahon Department of Computer Science, University of Miami will present Social Volunteer Computing While both volunteer computing and social networks have already proved to be successful, the merging of these two models is a new field. We call this new field Social Volunteer Computing. A Social Volunteer Computing system utilizes the relationships within a social network to determine how computational resources flow towards tasks that need to be completed. We call this Socially-Driven Computation. The result of this computation, which we call Socially-Computed Content, is added back into the social network. This presentation will also introduce RenderWeb, a prototype Social Volunteer Computing system that allows Blender animations to be quickly rendered by friends within Facebook. This is another in the Departmental Pizza Seminar series. Wednesday, January 27th, 2010 5:00 PM, Ungar Building, Room 402 Mr. Michael Goldberg Flamingo Software will present A Career in Software Development: Web-Based Systems Over 40 years ago, Michael Goldberg started a Miami software company called FDP, employing many UM mathematics and computer science students over the years and eventually becoming the leading provider of software for the insurance and pension industries. After selling FDP, Michael started another company 5 years ago called Flamingo Software, specializing in web-based systems for insurance and financial services companies. Find out what a career in software development can be like and what it takes to run a successful software company. This is a CSMS Project Industrial Liaison Seminar. Refreshments will be served. Wednesday, January 27th, 2010 2:30 PM, McArthur Engineering Annex, Room MEA202 PIRE: An Opportunity to Travel the World and to Participate in an International Research Collaboration The FIU School of Computing and Information Sciences, with partners FAU and UNCC, is leading an exciting international research and career development opportunity for students from FIU, FAU, UNCC, UM, and UPRM. This Partnership for International Research and Education (PIRE) is a 5-year long project funded for2.3 million through a highly competitive program by the National Science Foundation. PIRE aims to provide international research and training experiences to its participants by leveraging the established programs, resources, and community of the Latin American Grid. After the successful execution of this program in the first two years, we decided to open this program to all disciplines of Science and Engineering. Students selected by the PIRE program will become part of a prestigious international research network focusing on using cyberinfrastructure to solve challenging societal problems while training a generation of globally-oriented IT professionals who will become leaders in industry and academia. We are looking for highly qualified undergraduate, graduate, and post-doctoral students in Science and Engineering disciplines who are interested in working on critical real-world applications that use or extend cyberinfrastructure as part of an international R&D team. The PIRE program includes collaboration-driven trips (6 weeks to full semesters in length) to top research sites in Argentina, Brazil, China, France, India, Italy, Mexico, Spain, and more countries. PIRE will also sponsor fellowships of up to \$10,000 per semester and monetary awards for winners of PIRE annual poster sessions.

This is a joint presentation by the Department of Electrical and Computer Engineering and the Department of Computer Science.

Wednesday December 16th 2009 5:00pm, Ungar 411

Dr. Justinian Rosca
Siemens Corporate Research,

will present

Statistical Inference of Missing Speech Data in the ICA Domain

Over the last forty years, remarkable progress has been made in the area of speech separation and enhancement, however accurate estimation of clean speech in real-world environments is still a challenge. We address the problem of speech estimation as statistical estimation with "missing" data in the independent component analysis (ICA) domain. Missing components are substituted by values drawn from "similar" data in a multi-faceted ICA representation of the complete data. I will present the algorithm for the inference of missing data in the case of a fixed pattern of missing data, and then I will present an application of our approach to the problem of bandwidth extension, where speech is degraded by a fixed filtering process. I will show the capability of the algorithm to reconstruct fine missing details of the original data with little artifacts: information of the source signal can indeed be modeled and used in order to recreate a natural sounding source in adverse conditions. The extension of the method to statistical spectral inference according to random patterns of missingness promises progress in the long open problem of performing speech enhancement while enhancing the intelligibility of speech.

This work has been done in collaboration with Doru-Cristian Balcan (Carnegie Mellon University) and Timo Gerkmann (Bochum University).

Biography: Justinian Rosca is Program Manager in Audio, Signal Processing and Wireless Communications at Siemens Corporate Research in Princeton, USA. He is also Affiliate Professor, Department of Electrical Engineering of University of Washington, Seattle, USA. He received the Dipl. Eng. degree in Computers and Control Engineering from Bucharest Polytechnic University in 1984, the M.S. and Ph.D. degrees in Computer Science from University of Rochester in 1992 and 1997 respectively. Dr. Rosca is conducting research in statistical signal processing and radio management, with an emphasis on topics involving acquisition, management and processing of data with uncertainties, such as audio processing, blind signal separation, wireless management, adaptive principles in stochastic search and optimization, and probabilistic inference in artificial intelligence.

Tuesday December 8th 2009 5:00pm, Ungar TBA

Dr. Tom Murase
NEC Japan

will present

Resource Allocation for Cognitive Radio to Assure QoS

For new terminals such as Network PC and thin-client, broadband network access is becoming significant. For mobile users, wireless communication technology such as Cognitive Radio (CR) is expected to realize seamless communications. However, QoS consideration in CR has not been well investigated. This talk intorduces a QoS control framework which consists of such as optimization of resource allocation, and shows a result of "Geographical Optimal Route Selection with Cost-Constraint".

This is another in the Departmental Pizza Seminar series.

Wednesday December 2nd 2009 4:30pm, Ungar 411

CSC410/CSC411 Students
Department of Computer Science, University of Miami

will present their projects ...

Andres Acevedo - Planning of iPhone game development. This talk will be about the steps taken in the preparation and planning of iPhone game development. Additionally it will cover subjects such as hardware requirements to develop for the iPhone (and possible solutions), setting up projects using the Xcode SDK and Interface Builder, basic concepts of cocoa and cocoa touch, and the use of existent frameworks to assist game development.

Jose Hernandez - Axiom Selection for Automated Theorem Proving. Is Abraham Lincoln a mammal? Does there exist a capital which has the same latitude as Moscow that can get flooded? This talk will be on my work on a axiom selector mechanism (SInE) for Automated Theorem Provers. The axiom selector picks out the correct facts needed to solve a specific set of question. I will go over how I managed to get the Automated Theorem Provers integrated into the axiom selector; also problems that arose when I did and how I plan to solve them.

Alexandra Teyssandier - ATP with External Sources of Axioms for Dummies. The internet is a constantly expanding source of knowledge. External specifications allow ATPs to access information from web sources and databases to prove large theorems. The goal is to allow an end user to take a general knowledge, English conjecture and develop a conjecture in first order logic to answer questions such as "where is it sunny in the northern hemisphere today?" The first step is a simple scroll-list interface with searchable options that allows a user to find predicates relevant to their query. It links to external specifications from SUMO and XDB. Future plans include developing a more user-friendly, English language input.

Michael Schick - Changing the World of Theorem Proving: From Untyped to Typed. This talk will discuss logic languages, including First Order Form (FOF) and Typed First order Form (TFF). It will examine the TPTP and its uses, specifically related to Automated Theorem Proving. Additionally, it will show how to convert FOF axioms to TFF. Results generated from using TFF as compared to FOF, and the other potential advantages of TFF will be described.

Pascual Oliu - Fast Edit Distance and LCS Calculation Using Dominance Relationships. Calculating edit distances between two sequences is a fundamental problem in bioinformatics. A new algorithm computes edit distances efficiently by identifying dominance relationships in a dynamic programming matrix. This talk will discuss coding an efficient implementation of the algorithm in C and using it to solve the related LCS problem.

This is another in the Departmental Pizza Seminar series.

Tuesday November 24th 2009 5:00pm, Ungar 506

Saminda Abeyruwan
Department of Computer Science, University of Miami

will present

PrOnto: Unsupervised Lexico-Semantic Ontology Generation Using
Probabilistic Methods

An ontology is a formal, explicit specification of a conceptualization. The process of engineering an ontology for a domain using the top-down approach is a complex task that consumes a lot of time and effort (it's the knowledge acquisition bottleneck). When the domain is a substantially large amount of texts (corpus), one way to expedite this prior process is to reverse engineer the ontology from the corpus. In this talk I am presenting my master's research proposal on an efficient and effective method to solve the reverse engineering problem using lexico-semantic analysis and probabilistic reasoning.

This is another in the Departmental Pizza Seminar series.

Wednesday November 18th 2009 5:00pm, Ungar 411

Geoff Sutcliffe
Department of Computer Science, University of Miami

will present

How to Give a Successful Talk

Almost everyone has to give a public presentation at some time. Graduate students have to defend their theses, research students have to present their results, and many jobs require presentations. It is common to have little or no experience when you give your first presentation, and you may even be a little nervous! This talk is aimed at (graduate, research, and other) students. It describes how to present a successful talk, in a simple standard format. This talk does not try to teach general speaking skills, nor impose any personally preferred techniques. It covers the structure of a talk, the use of visual aids, speaking technique, and how to cope with questions.

This is another in the Departmental Pizza Seminar series.

Wednesday October 28th 2009 5:00pm, Ungar 411

Dr. Juan B. Gutierrez
Department of Mathematics, University of Miami

will present

Biostructural Classification Database (BCD) - Part 2

In the first part of this talk we defined the problem we want to solve, i.e. the need of a systematic and flexible approach to perform pattern classification of biological data coming from many sources. Since the nature of the problem dictates the nature of the solution, in this talk we will talk about the scope of the BCD, the limitations found during its implementation, and how they were resolved. The present state of the BCD and its future direction will be discussed. The bottom question is: how general can an information system be? As we will see, not much, i.e. 'very general' with a twist.

This is another in the Departmental Pizza Seminar series.

Wednesday, October 21st, 2009 5:00 PM, Ungar Building, Room 402

Prof. Tim Dixon
RSMAS/MGG, University of Miami

will present

Slow Earthquakes in the Costa Rica Subduction Zone

Subduction zones, where oceanic plates are pushed under the leading edge of continental plates along ocean trenches, produce Earth's largest earthquakes and most tsunamis. The pattern of strain release during earthquakes, where the leading edge of the continental plate jumps towards the ocean by several meters or more, is related to the slow build-up of strain accumulation during the interseismic period, which may last for hundreds of years. New GPS technology permits this slow pattern of strain accumulation to be measured. Studies of strain accumulation may give clues to the nature of future earthquakes, leading to improved understanding of the seismic process and improved forecast of seismic hazard. However, GPS data at a number of subuction zones indicates that not all accumulated strain is released during earthquakes; slow, aseismic slip events with durations of days - months are increasingly recognized as a major component in the strain release budget. In this talk I will describe a new GPS and seismic network that is being installed in northern Costa Rica by the University of Miami to monitor such events, and describe preliminary results from the first three years of operation. We have already observed one slow slip event, in May 2007. Maximum surface offsets were approximately 2 cm, occurring over a duration of several weeks, corresponding to an ~ M 6.5 earthquake if all of this strain had been released rapidly. Maximum slip was centered near the down-dip edge of the conventionally defined seismogenic zone. How these data are collected, analyzed and interpreted will be discussed, as well as implications for future earthquakes and improved understanding of the earthquake process.

This is a CSMS Project Industrial Liaison Seminar. Refreshments will be served.

Wednesday October 14th 2009 5:00pm, Ungar 411

Dr. Juan B. Gutierrez
Department of Mathematics, University of Miami

will present

Biostructural Classification Database (BCD)

Biodatabases are becoming increasingly accessible, increasingly big, and increasingly heterogeneous. The BCD is an (ongoing) open web information system that provides a systematic and flexible approach to perform pattern classification of biological data coming from many sources. In this talk I will describe the practical problems encountered during the design and development of the BCD and the solutions implemented related to high performance, high availability, high maintainability, high extensibility, cloud computing, database federation, web services, configuration management, and integration of third-party calculated feature vectors into the BCD data mining framework. The present state of the BCD and its future direction will be discussed. The bottom question is: how general can an information system be? As we will see, not much, i.e. "very general" with a twist.

This is another in the Departmental Pizza Seminar series.

Thursday, October 8th, 2009 5:00 PM, Ungar Building, Room 402

Dr. Arnaldo Horta
National Security Agency
Fort Meade, Maryland

will present

Inside the Puzzle Palace:
Careers in Mathematics and Computer Science
at the National Security Agency

In this talk, I will discuss the role of mathematics and computer science at NSA and discuss hiring opportunities, including REUs (Research Experiences for Undergraduates) and programs for graduate students.

This is a CSMS Project Industrial Liaison Seminar. Refreshments will be served.

Wednesday October 7th 2009 5:00pm, Ungar 411

Dr. Ubbo Visser
Department of Computer Science, University of Miami

will present

Statistics-based Real-time Sports Simulation

Autonomous agents in real-time and dynamic adversarial environments offer numerous research challenges. Perception, localization, decision- making, communication, and locomotion are good examples. The novel modern sports simulator we will discuss integrates results from ten years of research in the area of autonomous soccer playing robots (both softbots and physical robots) with RoboCup as a testbed. We will explore the problem of enabling autonomous agents in finding the right passing point or in making a complex decision within a soccer game while dealing with time constrains, hostile opponents, and dynamic environments. We propose a framework for spatio-temporal real-time analysis of dynamic scenes. The underlying hierarchical three-tier multiagent system consists of autonomous BDI agents that allows dynamic group structures (e.g., an emergent situation for a wing attack). The online game runs seamlessly in a web browser with a new and state-of-the-art 3D visualization engine. Latest developments include research results from a motion capturing lab and face generators to enhance the believability of the players and the users' visualization experience.

This is another in the Departmental Pizza Seminar series.

Wednesday September 30th 2009 5:00pm, Ungar 411

Saminda Abeyruwan
Department of Computer Science, University of Miami

will present

Advanced Web Services with Apache Axis2
Part 2 of "Web Services for Human Beings"

Web Services is one of the most dominant tools in the world today to implement SaaS (Software As A Service) principles. It has revolutionized the way services are mashed-up to build complex services and governing bodies. QoS (Quality of Services) such as security, reliability, etc., plays an important role in creating an ultimate service with governance. This presentation will focus on harnessing the power of Web Services using Apache Axis2 to create such a service and invoking prior service using Apache Axis2 client in a matter of minutes. In logic ...

! [X] : (superhuman(X) => (rocks(X) & knows_axis2_web_services(X)))

This is another in the Departmental Pizza Seminar series.

Wednesday September 23rd 2009 5:00pm, Ungar 411

Saminda Abeyruwan
Department of Computer Science, University of Miami

will present

Web Services for Human Beings - Apache Axis2

There is an enormous demand from the industry for Web Services and related technologies at present. These demands are not only from the IT field, but also from other disciplines such as finance, telecommunications, and government. Web Services is one of the most successful implementations of Service Oriented Architectures, which provides an infrastructure to build reusable software services in a heterogeneous environment. Apache Axis2 is considered to be the most successful open source Web Services middleware platform offered by the Apache Software Foundation. This presentation will focus on the basics of the Apache Axis2 architecture, concepts, and R&D opportunities available for students in the open source arena.

This is another in the Departmental Pizza Seminar series.

Wednesday September 2nd 2009 5:00pm, Ungar 411

Dr. Geoff Sutcliffe
Department of Computer Science, University of Miami

will present

External Sources of Axioms
in Automated Theorem Proving

(or ... "A Computer Program that can play Trivial Pursuit")

In recent years there has been a growing demand for Automated Theorem Proving (ATP) in large theories, which often have more axioms than can be handled effectively as normal internal axioms. This work addresses the issues of accessing external sources of axioms from a first-order logic ATP system, and presents an implemented ATP system that retrieves external axioms asynchronously, on demand. The efficacy of the system is demonstrated on test problems that make use of a range of external sources of axioms, including databases and web services. In the long term this system will be able to answer very general knowledge questions, using a natural language interface.

This is another in the Departmental Pizza Seminar series.

Thursday July 30th, 2009 2:00pm, Ungar Building, Room 426

Aman Kakkad
Department of Computer Science, University of Miami

will present

Machine Learning for Automated Theorem Proving

Developing logic and proving conjectures are some of the major goals in the world of Automated Theorem Proving. Various ATP systems have been designed for the purpose. However, for majority of ATP systems it becomes difficult to prove conjectures when the available axiom set is too large. This research introduces a concept of machine learning to deal with large axiom sets. The closest relevant axiom sets for a failed conjecture are formed by learning from existing solved conjectures in the same problem domain. Semantic axiom refining approaches are then applied on an already reduced axiom set for refining it to a much finer level. The failed conjecture is then tried proving from refined axiom set. The SWC problem domain was picked for testing purposes, and the tool proved successful by solving 24/124 failed conjectures. Performance was better when compared to SRASS.

This is a Department of Computer Science Masters Defense.

Wednesday, April 22th, 2009 5:00 PM, Ungar Building, Room 411

Tom Leffingwell
Network Architect
University of Miami

will present

UM Network Connectivity

Have you ever wondered how the University of Miami is connected to the rest of the world? In the past few years the University's network has grown to include four Internet Service Providers, 46 different commercial peers, and several research and education networks, including National Lamdba Rail, Internet 2, and others. Discover what equipment is used, what it looks like, and how it is connected together. Learn about what network resources are available to the University community.

A Pizza Science Seminar. Pizza and refershments are served.

Monday, April 13th, 2009 3:00 PM, Ungar Building, Room 426

Aravind Prakash
Computer Science, Univ of Miami

will present

Confidential Data Dispersion Using Thresholding

With growing trend in 'cloud computing' and increase in the data moving into the Internet, the need to store large amounts of data by service providers such as Google, Yahoo, Microsoft, etc. has increased over time. Now, more than ever, there is a need to efficiently and securely store large amounts of data. This thesis presents an implementation of a Ramp Scheme that confidentially splits a data file into a configurable number of parts or shares of equal size such that a subset of those shares can recover the data entirely. Furthermore, the implementation supports a threshold for data compromise and data checksum verification to verify that the data parts have not been tampered with. This thesis addresses two key problems faced in large-scale data storage, namely, data availability and confidentiality.

This is a Department of Computer Science Masters Defense.

Monday, April 13th, 2009 5:30 PM, Ungar Building, Room 426

Anoop Mohan
Computer Science, Univ of Miami

will present

Confidential and Resilient Store of Persistent Web Objects

Persistent and secure store for web objects is an attractive feature in today's web world and possess a good potential for exploration. Persistence of a storage mechanism refers to its ability to store an object for extremely long time periods. Resilience refers to its fault tolerance ability or its ability to retrieve the object completely even if a part of that object is lost due to any catastrophic failure like the disk failure. It is also important that this storage mechanism is able to store this object in a secure manner. In the current world usability of any storage mechanism is enhanced multiple times if it could be used from a web service. This thesis considers different techniques that provide each of these properties and proposes a storage mechanism that makes use of information dispersal techniques that is suited to store data securely with an emphasis on availability and resilience. A working prototype of this storage mechanism was developed as a part of this thesis and is made available as a library for program developers. This library provides APIs to store and retrieve data as well as a control API which can be used to query and set the configuration. The APIs to store and retrieve data also has a limited web interface which increases its usability to web developers. The performance of this prototype was measured and is presented using graphs and tables. Finally a demonstration of the applications of this prototype is also provided.

This is a Department of Computer Science Masters Defense.

Monday, April 6th, 2008 3:30 PM, Room MCA202

Dr. Frank Marks
Research Meteorologist and Director
NOAA/AOML Hurricane Research Division

will present

NOAA Hurricane Forecast Improvement Project (HFIP):
NOAA's 10-year plan to improve Hurricane Intensity Forecasts

Tropical cyclones continue to be a serious concern for the Nation, causing significant risk to life, property, and economic vitality. While NOAA has made steady improvements in forecasting track since 1990, intensity has lagged behind, primarily because a good track forecast is essential before addressing any intensity changes. However, over the last 5 years NOAA's track forecasts improved enough for us to now focus on intensity forecast improvements. Since the devastating 2004-2005 Hurricane Seasons NOAA developed a plan to improve forecasts of hurricane intensity, structure, and track in cooperation with other government and non-government partners. This Hurricane Forecasting Improvement Project (HFIP), is a 10-year project designed to accelerate improvements in one to five day forecasts for hurricane track, intensity, and structure and to reduce forecast uncertainty, with an emphasis on rapid intensity change because of its importance to emergency management amongst others. The HFIP Plan has been widely distributed and reviewed, with an alliance of the hurricane research and forecasting communities agreeing to move forward with its implementation. The Plan's strategy is to perform research into hurricane intensity change and rapid intensification mechanisms, exploit new observing systems, expand computing capacity, build higher resolution coupled models and ensembles, improve infrastructure for transitioning research to operations (R2O) and support operations-like systems to the research community (O2R), and broaden NOAA expertise and expand interaction with the external community.

This is a Department of Electrical and Computer Engineering presentation, and a CSMS Project Industrial Liaison Seminar. See Computer Science and Mathematics for Scientists. Refreshments will be served.

Wednesday, April 1st, 2009 5:00 PM, Ungar Building, Room 402

Neal Alewine, Harvey Ruback, and Juan Caraballo
IBM

will present

Speech Technology: Insights into the technology behind speech recognition and unique applications

Neal Alewine and Harvey Ruback will discuss IBM's speech technology. Also present will be the IBM LA Grid program director Juan Caraballo.

Neal Alewine is an IBM Senior Technical Staff Member, Software Group Voice and DataPower Architect. Harvey Ruback is an IBM Senior Technical Staff Member, Systems and Technology Group Advanced Voice Architect. Juan F. Caraballo is the IBM Program Director for LA Grid - Corporate University and Innovation Programs. He was previously the Software Group Program Director for Enterprise Speech Solutions.

This is a CSMS Project Industrial Liaison Seminar. See Computer Science and Mathematics for Scientists. Refreshments will be served.

Thursday, March 5, 2009 at 5:00pm, Ungar Building, Room 402

Professor Michael Mascagni

of the Department of Computer Science, Florida State University

presents

Serial and Parallel Random Number Generation: Theory and Practice

We will look at random number generation from the point-of-view of Monte Carlo computations. Thus, we will examine several serial methods of pseudorandom number generation and two different parallelization techniques. Among the techniques discussed with be "parameterization," which forms the basis for the Scalable Parallel Random Number Generators (SPRNG) library. SPRNG was developed several years ago by the author, and has become widely used within the international Monte Carlo community. SPRNG is briefly described, and the lecture ends with a short revue of quasirandom number generation. Quasirandom numbers offer many Monte Carlo applications the advantage of superior convergence rates.

Dr. Mascagni was born in Bologna, Italy. He has a Ph.D. in Mathematics from Courant Institute of Mathematics and has worked at the National Institute of Health and the Institute of Defense Analysis' Supercomputer Research Center. He is currently a professor in the department of Computer Science of Florida State University. Dr. Mascagni is on the editorial board of three journals in his field, and is on the Board of Directors of IMACS. He has been a visiting professor at the University of Padova in Italy, the University of Salzburg in Austria, and the Swiss Federal Technical Institute-Zurich in Switzerland, and is a consultant to industry and government

This is a joint presentation of The Center For Computational Science and the Department of Computer Science and co-sponsored by Rosenstiel School Of Marine And Atmospheric Science and the The Department Of Electrical And Computer Engineering

Friday, February 20th, 2009 at 4:00 pm, Ungar Building, Room 411

Joe Clarke

Cisco Systems

Presents

Porting GNOME to FreeBSD

Joe Clarke will discuss his role in porting GNOME/GTK+ to the FreeBSD platform. As part of the discussion, the goals, structure and charter of the GNOME Project will be outlined.

Joe Clarke works at Cisco Systems, and is an expert in CiscoWorks LAN Management Solution and IOS embedded device management. His day job supports his interests in FreeBSD, including his work as Ports Manager, GNOME porter, and Tinderbox maintainer. Presented by the Department of Computer Science

Refreshments will be served.

Friday, February 20, 2009 at 11:00; Ungar Room 411

Surajit Chaudhuri

Principal Researcher
Microsoft Research

will present

A Programming Framework for Data Cleaning

Data cleaning is a critical component of Business Intelligence software. Specifically, the data cleaning step reconciles multiple representations of the same data and thereby ensures accurate data analysis. With the advent of the web, this problem has gained even more importance. For example, data cleaning technology is also used to capture typos and differences in representation when looking up addresses in an online address search. Traditionally, data cleaning has been driven by consultants and software that is custom made for specific vertical domains. In this talk, we take a different approach and propose a programming framework for data cleaning. We will identify some of the key aspects of such a programming framework.

Surajit Chaudhuri is a Principal Researcher and a Research Area Manager at Microsoft Research, Redmond. In 1996, Surajit started the AutoAdmin project on self-tuning database systems at Microsoft Research and along with his team developed novel automated physical design tuning technology. Their research led to the development of Index Tuning Wizard (Microsoft SQL Server 7.0, SQL Server 2000) and Database Tuning Advisor (SQL Server 2005, SQL Server 2008). More recently, along with his colleagues, Surajit has worked in the area of data cleaning. Their research on data cleaning has been incorporated in Microsoft SQL Server 2005 and SQL Server 2008. Surajit is also interested in the problem of search and querying information exploiting IR as well as DBMS techniques. Surajit did his Ph.D. from Stanford University and worked at Hewlett-Packard Laboratories, Palo Alto from 1991-1995. He is an ACM Fellow. He was awarded the ACM SIGMOD Contributions award in 2004 and a 10 year VLDB Best paper Award in 2007.

Refreshments will be served. All are invited.

Thursday February 19 at 6:00 PM. Ungar 426.

Game Night at Computer Science

Video Game Society
of the
Department of Computer Science
Presents

Street Fighter 4
Launch Event

Everyone is welcome. Competition, Pizza, Chun Li.
For more information contact adam _at_ mail.cs.miami.edu.

Thursday February 19 at 6:00 PM. Ungar 426.

Game Night at Computer Science

Video Game Society
of the
Department of Computer Science
Presents

Street Fighter 4
Launch Event

Everyone is welcome. Competition, Pizza, Chun Li.
For more information contact adam _at_ mail.cs.miami.edu.

Wednesday, February 18th, 2009, 5:00pm, Ungar 411

Dr. Heinz Kierchhoff

CEO of Aitainment Gmbh, Germany

will present

New Generation Sports Games

Dr. Kierchhoff presents a new generation of e-Sports-games. It is a combination of characteristics of real soccer players with modern AI technology. The user adds soccer knowledge for the next game while choosing formation, tactics, and training sessions for his team. Each game is simulated and visualized in a 3D Java applet within a web browser thereafter. This results into a unique game that has convinced the German Bundesliga (DFL GmbH) among others. Dr. Kierchhoff is one of the driving forces behind the "Official Bundesliga Manager", a game that has convinced more than 105,000 users since August 2007 and is running on www.bundesliga.de.

Users can create their own team from all 530 Bundesliga players and plays with this team to win the championship against 17 other teams in a weekly rhythm (in analogy to the real Bundesliga). The game also contains options to ascend/descend into different leagues according to the quality of the team (virtual 1./2. Bundesliga). The games have the same rhythm than the reality, however, the games will be played one day earlier.

Modern technology is the basis for the game. The players are software robots and act autonomously. They decide on their own with the help of modern AI technology which action will be executed within the next few seconds. The combination of reality with weekly (daily possible) updates of player data, the definition of the user with his formation and tactics and the autonomy of the software robots is brand-new and guarantees a unique experience for users.

The Official Bundesliga Manager is a combination of high tech artificial intelligence methods, an new 3D graphics engine, and elite sports. Artificial software robots are charged with weekly updated real data from the Bundesliga players. They then play autonomously their league games in a virtual world. (www.bundesliga.de).

The talk contains a technology demonstration and will then give insight into a vibrant industry sector by analyzing more than 120.000 customers. Dr. Kierchhoff will talk about potential markets and business opportunities in this sector. The talk underlines the importance for game programming in education.

This is another in the Department of Computer Science Pizza Seminar Series.

Monday, February 16th, 2009 4:00 pm, Ungar Building, Room 411

Rendering Animations with Distributed Applets Over the Internet

Adam McMahon
Department of Computer Science
University of Miami

Thesis supervised by Professor Victor Milenkovic

High quality 3D rendering requires massive computing resources. In order to render animations within a reasonable amount of time, the rendering process is often distributed among a cluster of computers, typically called a rendering farm. However, most individuals and small studios do not have the resources to purchase or lease a rendering farm. In the late 1990's, Java technology brought a hope that distributed applets could be utilized as an alternative to traditional network rendering models. Yet, this hope was never realized, nor was it fully implemented. Taking into account new developments in web application technology and the Sunflow renderer, this thesis will reexamine the possibility of distributed rendering applets. This thesis will suggest that distributed Java applets can effectively render projects across a collection of heterogeneous and geographically dispersed computers over the Internet. Moreover, this paper will present a prototype web application, called RenderWeb, that uses distributed applets to quickly render projects created in popular animation programs, such as Blender, 3D Studio MAX, and Softimage

This is a Department of Computer Science Masters Defense

Thursday Jan 29 at 6:00 PM. Ungar 426.

Game Night at Computer Science

Video Game Society
of the
Department of Computer Science
Presents

Resident Evil 5
Prerelease Demonstration

Everyone is welcome. There will be games, pizza and zombies.

For more information contact adam _at_ mail.cs.miami.edu.

Thursday, December 4, 2008, 2:00pm, Ungar 411

Dr. Tom Murase

Principal Researcher
System Platforms Research Laboratories, NEC

will present

A Study On Cross-Layer Design For Qos Control In Wireless Lan

QoS (Quality of Service) control in WLAN (IEEE 802.11 Wireless LAN) is becoming increasingly important. Although previous research has attempted to increase total throughput, few studies have treated individual TCP/UDP flow QoS. EDCA in IEEE 802.11e might provide prioritized QoS functions that would partially address this problem. However, in uplink flow, which is defined as data moving from a terminal toward an Access Point, EDCA has limitations. These manifest themselves both across classes and in differentiated QoS control between terminals in the same class. Furthermore, 802.11e requires modification of terminals as well as other alterations proposed by other researchers. Instead of 802.11e or other modifications of 802.11, we propose an approach to controlling QoS that requires no terminal modifications or installation of additional software/hardware. The proposed idea is MAC-frame Receiving-Opportunity Control (ROC). Performance evaluation shows that the ROC causes some performance degradation in total WLAN throughput but can achieve not only QoS priority control but also arbitrary throughput performance. In particular, the ROC (in the MAC layer) can also permit different throughputs for high priority and low priority flows, conditioned on control processes in other layers. These may include rate adaptation (in the MAC layer) and TCP congestion control (in the TCP layer).

Wednesday, December 3, 2008, 5:30pm, Ungar 411

CSC410/CSC411 Project Students
Department of Computer Science
University of Miami

will give project presentations

---<>---

Internship Report - Andrew Phillip Mendola

Report on internship.

---<>---

Justin Stoecker and Richard Roesler
Department of Computer Science
University of Miami
Social Robotics and Health Care

As the aging baby boomer generation approaches retirement age, health care providers are searching for ways to ease the burden of an increasingly older population. Subsequent generations are migrating away from their parents and grandparents, increasing the relevance of assisted-living facilities for elderly patients. Moreover, a rising life expectancy and dependence on a greater number of health professionals is contributing to a more expensive health care system, and depression and loneliness is now common among seniors. Researchers in the field of social robotics propose the use of robots to assist caregivers in menial tasks and to extend the independence and quality of life of aging individuals where possible.

As part of Humana's Rapid Prototyping Internship Program, we developed a system to evaluate the potential for a robotic companion's use in the home. Pleo, a robotic dinosaur manufactured by UGOBE, serves as the platform used for this prototyping. Reprogramming Pleo to perform the roles of both a health care tool and companion may provide an interface to technology that is more approachable for the elderly. We present the details and challenges in building the software used to enhance Pleo and integrate him into a system of devices that work together. The processes involved in creating new media for the robot will also be covered. In addition, we will discuss the structure and modifiability of the Pleo robot, including its limitations and advantages.

This is another in the Departmental Pizza Seminar series.

Friday, November 21st, 2008 at 12:15 pm, Ungar Building, Room 426

Robust explicit construction of 3d configuration spaces using controlled linear perturbation

Steven Trac
Department of Computer Science
University of Miami

We present robust explicit construction of 3D configuration spaces using controlled linear perturbation. The input is two planar parts: a fixed set and a moving set, where each set is bounded by circle segments. The configuration space is the three-dimensional space of Euclidean transformation (translations plus rotations) of the moving set relative to the fixed set. The goal of constructing the 3D configuration space is to determine the boundary representation of the free space where the intersection of the moving set and fixed set is empty. To construct the configuration space, we use the controlled linear perturbation algorithm. The controlled linear perturbation algorithm assigns predicate polynomial signs that are correct for a nearly minimal input perturbation. The output of the algorithm is a consistent set of polynomial signs. This approach is algorithm-independent and overhead over traditional floating point methods is reasonable.

If the fixed and moving sets are computer representations of physical objects, then computing the configuration space greatly aids in many computational geometry problems. Our main focus of computing the configuration space is for the path planning problem. We must find if a path exists from the start to the goal, where the fixed set is the obstacle, and the moving set is the object trying to reach the goal.

This is a Department of Computer Science Ph.D. Defense

Thursday, November 13th, 2008 11:00 am, Ungar Building, Room 426

Aparna Yerikalapudi

Department of Computer Science
University of Miami

will present

Answer Extraction in Automated Reasoning

One aspect of Automated Reasoning deals with writing computer programs that can answer questions using logical reasoning. An Automated Theorem Proving (ATP) system translates a question to be answered to a first-order logic conjecture, and attempts to prove the conjecture from the set of axioms provided. If a proof is found an answer extraction method can be applied to answer the original question. If more than one proof is possible, more than one answer may need to be extracted. Most ATP systems can find only a single answer for a question. To extract multiple answers, users may have to re-invoke the ATP system with a modified question. This work describes some ways to achieve multiple-answer extraction. An answer extraction program, called the multi-answer system, has been developed to automate the process of answer extraction. It enables users to conveniently extract different answers in a variety of ways. Extracting more than one answer is useful but it is also helpful if the extracted answers can be compared to a list of user's answers. The multi-answer system uses set operations like union, intersection, subset, superset, element, and difference to perform answer comparison.

This is a Department of Computer Science Masters Defense

Wednesday, November 12th, 2008 5:00 PM, Ungar Building, Room 402

Dr. Peter Stephenson
Principal Scientist, Emrging Technology Application of the Innovation Center, Humana

and

Adam McMahon, Justin Stoecker, and Richard Roesler
Department of Computer Science, University of Miami

will present

How to save lives and influence people

With the number of economic, environmental and foreign policy crises at the moment, have you thought about how you as an individual can directly contribute? As a technologist, have you ever thought about how you could save a life, or even better, many? If you have, think about how to promote wellness in your community (health, environment, security and happiness). With all of the crises happening around us at the moment, this is one problem on which we as individuals can have a dramatic effect, for ourselves and those around us.

Within Humana, the Innovation Center is a group of business and technology leaders who are looking at how to engage people to improve their wellness. Within the Emerging Technology Applications group we use digital media and emerging technologies such as computer games, social media, virtual worlds, social robots and the like. We run a paid Rapid Prototyping Internship Program (RPIP) that is designed to provide students with the chance to temper their skills on problems that have important and real world applications.

In 2008, three University of Miami computer science students, Justin Stoecker (BSc 2009), Richard Roesler (BSc 2009) and Adam McMahon (MSc 2009) undertook internships. Adam used Sims2 to design a game to show the health and economic benefit of using technology to assist the elderly to stay in their homes when threatened with an assisted living facility. Justin and Richard developed a prototype system for using the Pleo social robot as an alternative interface for the elderly to use technology in the home such as medication reminders, wireless scales and blood pressure cuffs. Justin also developed a memory game for the Pleo.

In the seminar, Adam, Justin and Richard will present the results of their work. The details of the upcoming RPIP'09 will also be discussed. If you are interested in applying for a paid position, please come and see the type of projects we are interested in pursuing.

This is a CSMS Project Industrial Liaison Seminar. See Computer Science and Mathematics for Scientists. Refreshments will be served.

Wednesday, November 5th, 2008, 5:00pm, Ungar 411

Dipl.-Inf. Tobias Warden

TZI - Center for Computing Technologies, Intelligent Systems Department
Universitat Bremen

will present

From Robotic Soccer to Autonomous Logistics: Spatio-Temporal Real-Time Analysis of Dynamic Scenes and Multiagent Adaption in Dynamic Environments

The presentation in this pizza seminar will be split in two consecutive parts. The first presents research in the field of knowledge representation and high-level scene interpretation for autonomous agents in dynamic physical environments. The application context will be robotic soccer. To be more precise, the RoboCup 3D soccer simulation league where the University of Bremen participated with our Virtual Werder 3D agent team from 2004-2007.

After a brief introduction to the domain, I'll argue that a quantitative world model alone is inadequate for agents interacting in a highly dynamic physical environment such as (simulated) soccer. Without further efforts targeted at a hybrid knowledge base, the agents are ill-equipped to apprehend extended motion incidences. To meet this need, I propose a knowledge processing pipeline ranging from relevance-driven compilation of a concise, basic qualitative scene description to a knowledge-based detection of complex events, actions and action sequences. The detection is thereby conceived as a spatio-temporal pattern matching problem which can be solved using logic programming. A methodology for the formalization of motion patterns and their inner composition is introduced and consequently applied in a knowledge engineering process to capture human expertise about a large set of soccer-specific motion situations. The formalization scales in a natural way from low-level events towards strategic plays. I'll also propose an efficient knowledge-based detection strategy whose bottom-up search is based on the interrelationships among modeled patterns. Seizing a full-featured prototype, a thorough statistical evaluation of analysis quality and execution performance was conducted in realistic application scenarios. Results were promising with respect to precision and recall, using - as a baseline - precise, complete sensor input at a 5Hz update frequency. Also, the analysis is robust against noisy, incomplete input as encountered in application scenarios such as the RoboCup 3D Soccer Simulation League, howing a defensible graceful quality degradation. The approach is also shown suitable for non-restricted real-time employment.

The second part of the presentation will retain the focus on autonomous agent and multi-agent systems. Yet, as it is motivated by my current occupation as a research assistant in the Collaborative Research Center entitled 'Autonomous Cooperating Logistic Processes', the application domain will change to transport logistics.

The CRC is concerned with research on new paradigms that can replace and effectively optimize traditional centralized management and control of complex supply networks in emergent, fast-moving, globalized markets. The basic idea pursued in our research group is a distribution of control and thus a transfer of autonomy from the human controller to the logistic objects themselves. These objects, which might be means of transport, freight containers or (collections of) bulk goods, are thereby modeled as intelligent software agents which pro-actively assume responsibility for transport logistic tasks such as the efficient and adaptive routing of goods from the production facilities in far east to the end user market in Germany.

I will try and sketch current tentative ideas to provide heterogeneous teams of agents which can be collectively thought of as acting as a freight forwarding company could adapt to a dynamically changing logistic environment via a collaborative multi-agent learning process. The adaption is conceived as running concurrently to the primary logistic task of the agent community. Depending on their individual circumstances/capabilities (mobile, resource-bounded agents due to live on on-board embedded systems or stationary high-performance agents, located at transshipment centers), the distinct agent types contribute to the global task of team adaption in several ways. For instance, as experience canvassers or as local learners/knowledge miners. The concept strives to bring together in a beneficial way research from multi-agent systems and multi-agent-based simulation, knowledge representation, learning and data/association rule mining.

This part of the talk is intentionally presented at an early stage of concept formation in order to draw comments and ideas from the audience.

This is another in the Department of Computer Science Pizza Seminar series

Wednesday, October 29th 2008 5:00pm, Ungar 411

Dr. Ubbo Visser

Department of Computer Science,
University of Miami

will present

AI-Simulation and 3D Visualization: New Generation Soccer Manager Games

Dr. Visser presents a new generation of e-Sports-games. It is a combination of characteristics of real soccer players with modern AI technology. The user adds soccer knowledge for the next game while choosing formation, tactics, and training sessions for his team. Each game is simulated and visualized in a 3D Java applet within a web browser thereafter. This results into a unique game that has convinced the German Bundesliga (DFL GmbH) among others. Dr. Visser is one of the driving forces behind the "Official Bundesliga Manager", a game that has convinced nearly 100.000 users since August 2007 and is running on www.bundesliga.de. Users can create their own team from all 530 Bundesliga players and plays with this team to win the championship against 17 other teams in a weekly rhythm (in analogy to the real Bundesliga). The game also contains options to ascend/descend into different leagues according to the quality of the team (virtual 1./2. Bundesliga). The games have the same rhythm than the reality, however, the games will be played one day earlier.

Modern technology is the basis for the game. The players are software robots and act autonomously. They decide on their own with the help of modern AI technology which action will be executed within the next few seconds. The combination of reality with weekly (daily possible) updates of player data, the definition of the user with his formation and tactics and the autonomy of the software robots is brand-new and guarantees a unique experience for users. The Official Bundesliga Manager is a combination of high tech artificial intelligence methods, an new 3D graphics engine, and elite sports. Artificial software robots are charged with weekly updated real data from the Bundesliga players. They then play autonomously their league games in a virtual world. (www.bundesliga.de). The talk contains details about the technology, an online demonstration, and an outlook within the context of the Graphics and Games Track at UM.

Ubbo Visser is currently a Visiting Associate Professor at the Department of Computer Science since August 2008. He received his Habilitation in Computer Science (qualification for Professor) from University of Bremen in 2003, a PhD in Geoinformatics from University of Muenster in 1995, and a MSc in Geography (Landscape-ecology) from University of Muenster in 1988. His research specialization is in artificial intelligence, more specifically on knowledge representation and reasoning. He is interested in the combination of symbolic and sub-symbolic technologies in the domain areas of "Semantic Web" and "Multiagent Systems". His focus in the Semantic Web area lies in the development of methods that combine terminological logics and spatio-temporal representation and reasoning techniques. The focus in the Multiagent Systems area lies in the development of techniques for agents that act in highly dynamic and real-time environments, such as mobile robots and games.

This is another in the Department of Computer Science Pizza Seminar series

Wednesday, October 22th 2008 5:00pm, Ungar 411

Dr. Fei Wang
School of Computing & Information Sciences,
Florida International University

will present

Graph Based Semi-supervised Learning

Graph based semi-supervised learning (GBSSL) has attached considerable interests in machine learning, data mining and computer vision communities in recent years. In this talk, I will introduce the basic configurations in GBSSL, and a representative algorithm -- label propagation, and a multilevel scheme to make it more efficient. The applications of GBSSL in image segmentation, text classification and information retrieval will also be introduced in this talk.

Dr. Fei Wang has got his Ph.d. degree from Tsinghua University, Beijing, China in 2008. He is now a postdoctoral researcher in School of Computing & Information Sciences, Florida International University. His main research interests include machine learning, data mining, information retrieval and pattern recognition. He has published over 40 papers in the top journal & conferences of the relevant field.

This is another in the Department of Computer Science Pizza Seminar series

Wednesday, October 15th 2008 5:00pm, Ungar 402

Donna A. Flores
Patent Attorney
Christopher & Weisberg, P.A.

will present

Intellectual Property Law: A Career Choice for Computer Science, Mathematics, and Engineering Majors

A brief overview of Intellectual Property (IP) Law including patents, trademarks, and copyrights. What is Intellectual Property? What role does an IP attorney play? Why is an engineering or scientific degree needed to become a patent attorney? Specific focus aimed at the process of becoming a patent attorney or agent, as well as the benefits and obstacles of pursuing this career choice.

This is a CSMS Project Industrial Liaison Seminar. See Computer Science and Mathematics for Scientists. Refreshments will be served.

Monday, October 13th 2008 10:00am, Engineering Building Annex, MEA202

Dr. Dana Ballard and Dr. Mary Hayhoe
University of Texas, Austin, TX

will present

THE ROLE OF EYE FIXATIONS IN THE COGNITIVE DIRECTION OF NATURAL TASKS

A human cognitive model that can robustly execute basic tasks would make a valuable contribution to three interrelated areas of endeavor, robust agent simulations, human-computer interfaces, models of human cognition. The centerpiece of the human execution of natural tasks is the fluid use of gaze to direct and select appropriate behaviors, but little is known about the control mechanisms responsible for the gaze sequences that do this. For example, a difficult problem for all models of gaze control, intrinsic to selective perceptual systems, is how to detect important stimuli without consuming excessive computational resources. We show in both real and virtual walking environments, that human gaze patterns are tightly controlled by task demands and remarkably sensitive to the probabilistic structure of the environment. One way to understanding this use of gaze is to create a model of a human that has a sufficient amount of complexity so as to be capable of generating such behaviors. Recent technological advances in graphics models that simulate extensive human capabilities can be used as platforms from which to develop synthetic models of visuo-motor behavior. Currently such models can capture only a small portion of a full behavioral repertoire, but for the behaviors that they do model, they can describe complete visuo-motor subsystems at a useful level of detail. The value in doing so is that the body's elaborate visuo-motor structures greatly simplify the specification of the abstract behaviors that guide them. The net result is that, essentially, one is faced with proposing an embodied "operating system" model for picking the right set of abstract behaviors at each instant. We outline one such model. A centerpiece of the model uses vision to aid the behavior that has the most to gain from taking environmental measurements. Preliminary tests of the model against human performance in realistic VR environments show that main features of the model show up in human behavior.

Colloquium sponsored by the University of Miami Departments of Computer Science and of Electrical and Computer Engineering and the Center for Computational Science

Wednesday, October 8th 2008 5:00pm, Ungar 411

Dr. Rahul Tripathi
Computer Science and Engineering
University of South Florida

will present

An Application of Randomness: Locally Random Reductions and Private Information Retrieval Schemes

I will talk about the problem of using shared resources for private computations. This problem has been studied in two closely related models: locally random reductions (LRRs) and private information retrieval (PIR) schemes.

In both models, there are two parties that communicate with each other. The first party consists only of a single client and the other party consists of one or more resources (which could be, for instance, oracles or servers). The client wishes to solve some problem on an input that is private to the client by interacting with the information resources. The crucial privacy requirement is that at the end of all communication each resource must not learn anything about the actual input, except possibly the length of the input. It is considered infeasible for the resources to simply send all information (or computational aids) that would be required by the client to solve the problem on any possible input of a given length.

The solution to this puzzle makes use of randomness in a crucial way. Basically, for any given input the client sends different random queries to different resources and uses the answers to these queries to solve the problem. The probability distribution of the random query(ies) sent to a single resource depends only on the length of the input. If we assume that the resources do not collude to learn the input of the client and that thus each resource gets to know only its own queries, then the resources do not learn anything about the actual problem instance. The challenge in constructing good LRRs and PIR schemes is to devise mechanisms with as low as possible communication requirement.

I will explain and motivate these models, describe known results related to these models, and present my recent results on the complexity of languages that have locally random reductions (LRRs) of certain types.

About the speaker: Dr. Rahul Tripathi is an assistant professor of Computer Science and Engineering at the University of South Florida (USF), Tampa, FL. His research interests are in algorithms and computational complexity theory. Dr. Tripathi holds a B.Tech. (2000) in Computer Science and Engineering from the Indian Institute of Technology, Kanpur, India, and an M.S. (2003) and Ph.D. (2005) in Computer Science from the University of Rochester, NY.

This is another in the Department of Computer Science Pizza Seminar series

Wednesday, October 1st 2008 5:00pm, Ungar 411

Zac Brown
Department of Computer Science
University of Miami

will present

Life of a Google Intern

Zac Brown, an undergraduate student in the University of Miami's Department of Computer Science recounts his experience as an intern with the world's most well known search engine and internet ad provider, Google. He will discuss his time there, the project he worked on and provide some insight into what Google is doing, where Google is going and some fun facts about Google. He may even provide confirmation to any rumors people have heard, provided his NDA permits him to do so. If you've ever wondered what it's like to work at Google, how they develop software, or even just how they make money, this is your chance to talk to a past Google employee.

This is another in the Department of Computer Science Pizza Seminar series

Thursday September 22nd 2008 5:00pm, Ungar 426

Dr. Carolina Cruz-Neira
Louisiana Immersive Technologies Enterprise (LITE) and
University of Louisiana at Lafayette

will present

A Fresh Look to Academic and Research Opportunities in
Visualization and Virtual Reality

Virtual environments have the way we understand and interact with data by transforming it into "in-our-face-and-hands" information. These environments have also revolutionized the way we communicate that information to others. We are now able to gain insight in particular problem domains like never before: we can collaborate with our colleagues, locally or remotely, and we can better explain our discoveries to others through powerful virtual reality experiences. Today, the practical implementation of these capabilities is still evolving, but a growing interest and acceptance by industry is accelerating the development efforts. The non-academic communities are demanding virtual reality and virtual environments technology that is robust and stable enough to come out of research laboratories and be put to work on industrial settings. This, combined with today's limitations of traditional sources of academic funding, presents a unique opportunity for universities open to widen their academic and research models towards applied research and entrepreneurship of their faculty and students. This talk reflects Dr. Cruz's experiences and personal assessment on the current state and trends of virtual reality technology, her challenging career as a non-traditional faculty, and the opportunities to develop strong integrated academic and research programs to capitalize in this emerging field.

Dr. Carolina Cruz-Neira is a pioneer in the field of Virtual Reality (VR). She started in the early 90's being the lead designer of the CAVE Virtual Reality Environment, and the software engineer of CAVELibs, the first software API to run such complex virtual environments. Until 2005, Dr. Cruz was the Stanley Chair in Interdisciplinary Engineering and the Associate Director and co-founder of the Virtual Reality Applications Center at Iowa State University (ISU). Currently, Dr. Cruz is the Executive Director and Chief Scientist of the Louisiana Immersive Technologies Enterprise (LITE) and the William Hansen Hall Endowed Chair in Computer Engineering at the University of Louisiana at Lafayette.

Colloquium sponsored by the University of Miami Department of Computer Science and the Center for Computational Science.

Wednesday, April 23rd 2008, 5:00pm, Ungar 411

CSC410/CSC411 Project Students
Department of Computer Science
University of Miami

will give project presentations

Become a Game Creator by Dragging and Clicking - Tommy Reddish

The focus of my project is to enable anyone to develop a simple to intermediate version of popular game genres without having to produce code or logic. The Easy GameCreator only requires the user to click or drag objects onto the game he/she is creating. In assisting the user to produce a game, the E.G.C. will organize the various stages and object files that the user chooses and displays them in a convenient location.

Mosquito Simulation - Geoffrey Sholler

This project, programmed in JAVA using the Processing graphics library, offers a visual simulation of the movements and interactions of mosquitoes with respect to humans. The inputs include the size of the space, mosquito and human population, and mosquito lifetime. The output consists of representative markers designated as humans or mosquitoes by different size and color. How often each person is bitten will be recorded as a counted number and as a gradual change in color of the human. This simulation should give a better understanding of how mosquitoes move and interact.

A Simple Facebook Wishlist - Ben Li

Ruby On Rails is a full stack MVC (model-view-controller) web framework that is designed for rapid application development. The popular Facebook application provides an API to allow third parties to write web applications that can be integrated with the Facebook web application itself. Amazon Web Services is a technology platform provided by Amazon.com. It allows developers to deploy web applications on a "cloud" of computing resources that is easily scalable. The Google AJAX Search API lets developers use JavaScript to embed a simple, dynamic Google search box and display search results. I wrote a simple facebook wishlist application that allows the user to search and display a wishlist of products on their profile, using with the Ruby on Rails framework, Amazon Web Services and Google Search APIs.

Conference Registration Software - Tiffany Fleming

This project is to design a software for conference registration. Original software was started as a project for Software engineering class. More universal and platform independent design is used this time around. My SQL was a software of choice. I would like to create a system with a semi-professional feel to it. Dr .U. Sarkar approved the project, though I had only minimal interaction with him; much of it will be left up to my ability and logic in finding the right way to manipulate the PHP and MySQL languages. It will be nice to implement the entire project in the web environment.

This is another in the Departmental Pizza Seminar series.

Thursday April 17th 2008 5:00pm, Ungar 506

Prof. Michael Mascagni
Department of Computer Science
Florida State University

will present

Random Number Generation: A Practitioner's Overview

We will look at random number generation from the point-of-view of Monte Carlo computations. Thus, we will examine several serial methods of pseudorandom number generation and two different parallelization techniques. Among the techniques discussed with be 'parameterization,' which forms the basis for the Scalable Parallel Random Number Generators (SPRNG) library. SPRNG was developed several years ago by the author, and has become widely used within the international Monte Carlo community. SPRNG is briefly described, and the lecture ends with a short revue of quasirandom number generation. Quasirandom numbers offer many Monte Carlo applications the advantage of superior convergence rates.

Dr. Mascagni was born in Bologna, Italy. He has a Ph.D. in Mathematics from Courant Institute of Mathematics and has worked at the National Institute of Health and the Institute of Defense Analysis's Supercomputer Research Center. He has currently a professor in the department of Computer Science of Florida State University. Dr. Mascagni is on the editorial board of three journals in his field, is on Board of Directors of IMACS. He has been a visiting professor at the University of Padova in Italy, the University of Salzburg in Austria, and the Swiss Federal Technical Institute-Zurich in Switzerland, and is a consultant to industry and government.

Wednesday, April 16th 2008, 5:00pm, Ungar 411

CSC410/CSC411 Project Students
Department of Computer Science
University of Miami

will give project presentations

Utilizing Latent Semantics to Determine Axiom Relevance - Alex Roederer

It is essential that a Theorem Prover, presented with a theorem to prove and a list of axioms to prove it with, knows which of the axioms will be most useful, and which will be of least use in its quest. The Google Prophet utilizes the power of latent semantics (the idea that any symbol has a "hidden" meaning that can be inferred by the other symbols it associates with) to rank axioms in order of most useful to least useful, with an accuracy far beyond simple string matching methods.

This is another in the Departmental Pizza Seminar series.

Wednesday, April 9th 2008, 5:00pm, Ungar 411

Prof. Erol Gelenbe
Electrical and Electronic Engineering Department
Imperial College London

will present

Optimised Route Search and Travel Time in
Uncertain Wired and Wireless Networks

We will present both an experimental framework for the study of on-line optimisation algorithms for packet routing, and theoretical results on "best paths" and on travel times in random network environments. The experimental "cognitive packet network" (CPN) test-bed at Imperial College will be described, and the CPN routing algorithm will be presented together with detailed experimental results concerning its performance. We shall also show how Brownian motion can be used to predict travel delays in uncertain wired and wireless networks, and describe an application to adaptive routing in wireless sensor networks. This work is currently funded by the US/UK International Technology Alliance funded by DoD and IBM.

Prof Erol Gelenbe FIEEE, FACM, FIEE, is the Professor in the Dennis Gabor Chair at Imperial College. A graduate of the Middle East Technical University, Ankara, Turkey, he is an elected Member of Academia Europaea and of the Turkish Academy of Sciences, and has received honoris causa doctorates from the University of Rome, the University of Liege (Belgium), and Bogazici University (Turkey). His research spans both computer networks and bio-informatics. A Member of Eta Kappa Nu, he has been decorated by the President of Italy with the honours of "Grand Officer of the Order of the Star of Italian Solidarity" and of "Commander of Merit of the Republic of Italy". In France he has received the honour of Officer of the Order of Merit, and has won the Grand Prix France Telecom of the French Academy of Sciences. Prof Gelenbe is on the editorial board of the Proceedings of the Royal Society, Acta Informatica, Telecommunication Systems and several other journals; he is Editor-in-Chief of The Computer Journal (British Computer Society).

This is another in the Departmental Pizza Seminar series.

Wednesday, April 2th 2008, 5:00pm, Ungar 411

Geoff Sutcliffe
Department of Computer Science
University of Miami

will present

How to Give a Successful Talk

Almost everyone has to give a public presentation at some time. Graduate students have to defend their theses, research students have to present their results, and many jobs require presentations. It is common to have little or no experience when you give your first presentation, and you may even be a little nervous! This talk is aimed at (graduate, research, and other) students. It describes how to give a successful talk, in a simple standard format. This talk does not try to teach general speaking skills, nor impose any personally preferred techniques. It covers the structure of a talk, the use of visual aids, speaking technique, and how to cope with questions.

This is another in the Departmental Pizza Seminar series.

Wednesday March 5th 2008 5:00pm, Ungar 402

Dr. Rong Jin
Department of Computer Science and Engineering
Michigan State University

will present

BoostCluster: Boosting Clustering by Pairwise Constraints

Data clustering is an important task in many disciplines. A large number of studies have attempted to improve clustering by using the side information that is often encoded as pairwise constraints. However, these studies focus on designing special clustering algorithms that can effectively exploit the pairwise constraints. We present a boosting framework for data clustering, termed as BoostCluster, that is able to iteratively improve the accuracy of any given clustering algorithm by exploiting the pairwise constraints. The key challenge in designing a boosting framework for data clustering is how to influence an arbitrary clustering algorithm with the side information since clustering algorithms by definition are unsupervised. The proposed framework addresses this problem by dynamically generating new data representations at each iteration that are, on the one hand, adapted to the clustering results at previous iterations by the given algorithm, and on the other hand consistent with the given side information. Our empirical study shows that the proposed boosting framework is effective in improving the performance of a number of popular clustering algorithms (K-means, partitional SingleLink, spectral clustering), and its performance is comparable to the state-of-the-art algorithms for data clustering with side information.

Dr. Rong Jin is an assistant professor of Computer Science and Engineering at Michigan State University. He is conducting research in the areas of statistical machine learning and its application to information retrieval. He has published over eighty research articles. Dr. Jin holds a B.A. in Engineering from Tianjin University in China, an M.S. in Physics from Beijing University in China, and an M.S. and a Ph.D. in Computer Science from Carnegie Mellon University. He received the NSF CAREER Award in 2006.

This is another in the Departmental Pizza Seminar series.

Monday March 3rd 2008 12:00noon, Ungar 411

Ziya Arnavut
Department of Computer Science
SUNY Fredonia

will present

Lossless ECG Signal Compression

Many transform-based compression techniques have been investigated and devised for ECG (Electrocardiogram) signal compression. However, the Burrows-Wheeler Transformation has not been completely investigated. In this talk, we show that when compressing ECG signals, utilization of linear prediction, Burrows-Wheeler Transformation, and inversion ranks yield better compression gain in terms of weighted average bit per sample than recently proposed ECG-specific coders.

Monday March 3rd 2008 5:15pm, Ungar 402 (Refreshments at 5:00pm)

Mr. Jonathan Jannarone
Vice President and Actuary
Assurant Group

will present

Actuaries in the workplace: Solving math puzzles for a living

Actuaries help insurance companies in many areas, such as accounting, claims, lawsuits, investments, mergers & acquisitions, and premium calculations. Join Jonathan Jannarone, UM math graduate and Vice President of Life Actuarial at Assurant, as he explains how actuaries apply math, statistics, and computer science to their daily job. Learn about how to become an actuary, consistently picked as one of the top five occupations in the country.

This is a CSMS Project colloquium

Wednesday February 13th 2008 5:00pm, Ungar 411

Dr. L. Miguel Encarnação, Dr. Peter Stevenson
Humana Inc.

will present

New Opportunites for Research in Health Services:
Games for Health and Visual Analytics

As governments and the health care industry struggle with the spiraling cost of health care, proactive approaches such as preventative care, early intervention and treatment adherence procedures are necessary if we are to reign in ballooning expenditure and protect ourselves against epidemic diseases. As we face new waves of long term debilitating diseases in the developed world such as obesity, diabetes, Alzheimer's and cardio-vascular disease, our ability to identify those at risk, to intervene, educate and change behavior, and to assist the affected to maintain their treatment regiments will become vital to keeping people and our economy healthy. In this seminar, we will discuss two programs being developed ay Humana, Inc. The Visual Analytics program seeks to give analysts the ability to analyze, correlated with external data sources, and report on the massive amounts of data collected on our member population in an interactive environment. New tools, techniques and technologies are being explored to identifying patterns, trends and at-risk groups to focus resources and intervention programs. The Games for Health program promotes understanding, fitness and cognitive maintenance through the use of computer games, exploring this technology as a new communication medium.

This is another in the Departmental Pizza Seminar series.

Friday February 8th 2008 2:30pm, Ungar 402

Dr. Wagner Meira Jr.
Computer Science Department
Universidade Federal de Minas Gerais, Brazil

will present

Scalability and Efficiency on Data Mining Applied to Internet Applications

The Internet went well beyond a technology artifact, increasingly becoming a social interaction tool. These interactions are usually complex and hard to analyze automatically, demanding the research and development of novel data mining techniques that handle the individual characteristics of each application scenario. Notice that these data mining techniques, similarly to other machine learning techniques, are intensive in terms of both computation and I/O, motivating the development of new paradigms, programming environments, and parallel algorithms that support scalable and efficient applications. In this talk we present some results that justify not only the need for developing these new techniques, as well as their parallelization.

Wagner Meira Jr. obtained his PhD from the University of Rochester in 1997 and is currently an Associate Professor at the Computer Science Department at Universidade Federal de Minas Gerais, Brazil. His research focuses on scalability and efficiency of large scale parallel and distributed systems, from massively parallel to Internet-based platforms, and on data mining algorithms, their parallelization, and application to areas such as information retrieval, bioinformatics, and e-governance.

Monday February 4th 2008 8:00pm, Cox 126

Professor Daniel Huttenlocher
Cornell University
(Phi Beta Kappa Visiting Scholar)

will present

Computational Social Science:
Large-scale studies of Wikis, Blogs, Social Networking

Many social interactions that are ephemeral in the physical world are recorded and accessible in the online world. The widespread use of online systems such as blogs, wikis and social networking sites provides a treasure trove of information about human behavior. This talk discusses some recent studies of large-scale online social systems, and speculates about what these studies suggest about human interactions more generally.

There will be a reception following the lecture. This colloquium is free and open to the public.

Wednesday November 28th 2007 5:00pm, Ungar 411 (Refreshments at 4:30pm)

Dr. Dimitris Papamichail
Department of Computer Science
University of Miami

will present

Opportunities at the borders of Biology, Computer Science and Mathematics

Starting in the 1990s, a large number of young scientists left the academic life for jobs with a score of companies that wanted to merge biology with computer science to create a new branch of science - bioinformatics. Accomplishments, such as the determination of sequences of the 3 billion chemical base pairs that make up human DNA, were realized through the combined efforts of life and computer scientists.

Flesh and bones start as protein. Inside each cell is a sort of instruction book that tells the cellular machinery what type of protein to produce and, more importantly, when. That instruction book is written with a 4-letter alphabet: A, C, G, T. Reducing life to DNA reduces life to numbers. And once a problem can be stated numerically, it can be mathematically modeled, stored in computer databases and analyzed with a variety of software tools.

Today, more than even, the field of bioinformatics offers exceptional opportunities for scientists that want to contribute in understanding the natural world. Outstanding challenges, including gene regulation and function, protein synthesis, interaction and conservation, disease-susceptibility prediction based on gene sequence variation and even evolutionary conservation, just to name a few, are all areas where major advances will happen in silico as much as in vitro. And this fact is acknowledged by both academic and industrial forces, which are currently recruiting and offering a variety of well rewarded positions to qualified researchers bridging the cultural worlds of life and mathematical sciences.

This is a CSMS Project colloquium

Wednesday October 10th 2007 5:00pm, Ungar 411 (Refreshments at 4:30pm)

Johannes Gudmundsson
Chief Executive Officer
iNECTA LLC

will present

Implementing and Maintaining IT Infrastructure and Business Software
for Mid-size Corporations - A VARs Perspective

Modern markets require businesses to stay on the edge of technology. At the same time, C-Level executives are demanding lower IT budgets. Management is forced to downsize their in-house IT departments and outsource the day to day maintenance and configuration to Value Added Resellers (VARs) . This shift towards greater reliance on external resources has made the VAR role increasingly more demanding and complicated, as it covers virtually every aspect of a corporation's IT need. In this talk, I will describe the current landscape for IT VARs and how technology is changing the workplace. Agenda: Segmenting the market space; the IT layers of a mid-size corporation; major players in each layer, in particular Microsoft, CISCO and Citrix; VAR survival skills; Conclusion.

This is a CSMS Project colloquium

Monday October 8th 2007 5:00pm, Ungar 402

Dr. Ubbo Visser
TZI - Center for Computing Technologies
Universität Bremen

will present

Soccer-playing robots: Solving Problems in Real-time, Dynamic, and Adversarial Environments

Autonomous mobile robots with the capacity to perform specific tasks such as monitoring security areas or assist in rescue scenarios have become increasingly important. Preparing one robot to have these abilities involves a variety of research questions. How does the robot know where it is? How does the robot get from A to B? What kind of action does the robot choose if an unforeseen situation arises? What is the best locomotion algorithm? Can the robot predict situations and, in general how can we measure the performance of these robots? The talk comprises a summary of a number of research projects that we have carried out at the University of Bremen. We will present problems, research approaches and solutions tackling localization, navigation, decision-making, bi-ped walking as well as situation recognition and prediction. We present how we have explored these issues on both robotic and software agents, and show examples of solutions arrived at within the context of soccer playing robots. Indeed, RoboCup has been fostering research on these problems by organizing competitions yearly, in a similar spirit that chess competitions had been used in the past decades to provide a performance measure of the Artificial Intelligence of the system. Because the soccer environment is adversarial with constant changes it demands for adequate approaches for a team of robots (rather than for one single robot). It provides a platform to evaluate how the research questions described in this talk can be explored on multi agents that communicate and cooperate. We then discuss how these results are relevant to other domains, and how they can be used and generalized in other domains such as the traffic domain or the medical domain.

Ubbo Visser is a Privat-Dozent (Assistant Professor) at the Department of Mathematics and Computer Science, University of Bremen, Germany since 1999. He received his Habilitation in Computer Science (qualification for Professor) from University of Bremen in 2003, a PhD in Geoinformatics from University of Muenster in 1995, and a MSc in Geography (Landscape-ecology) from University of Muenster in 1988. His research specialization is in artificial intelligence, more specifically on knowledge representation and reasoning. He is interested in the combination of symbolic and sub-symbolic technologies in the domain areas of "Semantic Web" and "Multiagent Systems". His focus in the Semantic Web area lies in the development of methods that combine terminological logics and spatio-temporal representation and reasoning techniques. The focus in the Multiagent Systems area lies in the development of techniques for agents that act in highly dynamic and real-time environments, such as mobile robots. Dr. Visser has published over 75 scientific articles and has received public research funding at international (Australia, European Commission), national (Germany), and regional (Bremen State) levels - including DFG (German Research Council), BMBF (Federal Ministry for Education and Research), ARC (Australian Research Council), and also from Industry. He has chaired a variety of international conferences, workshops, tutorials and symposia (e.g. RoboCup, IJCAI- and ECAI-Workshops) and he has won multiple prizes and awards for outstanding research papers and software engineering (e.g. best artificial intelligence application in Germany, awarded by the German AI society). He is the Co-Founder of three software companies that develop products based on modern AI-technologies (Conterra Ltd., ProPlant Ltd., Coach and Win Ltd.). He is a trustee of the RoboCup Federation and an Executive Member of the Media & IT Association of the state of Bremen. He is the Co- Editor of the German Journal of Artificial Intelligence and member of the editorial board of the International Journal on Semantic Web and Information Systems.

Wednesday September 26th 2007 5:00pm, MEA 202

Dr. Stephan Schulz
Comsoft GmbH
Karlsruhe, Germany

will present

ADS-B: Surveillance for Cooperative Air Traffic

Both the increase in civilian air traffic and the desire for improved routing and scheduling of aircraft put larger and larger demands on air traffic control (ATC). Automatic Dependent Surveillance - Broadcast (ADS-B) is a technique developed to meet the resulting demand for better information about the air situation. Civilian aircraft have a strong desire to be visible to ATC, and are willing to contribute to that end. This is already used in secondary radar, where aircraft transponders actively reply to radar requests and provide air traffic control with additional information like the aircraft identity and altitude. ADS-B delegates much more responsibility to the aircraft. The aircraft periodically determines its own position using a satellite-based navigation system, and broadcasts this information. The ADS-B ground station is a passive device that receives and decodes this signal, and forwards the information to ATC. ADS-B sensors can be cheap, compact, and only have minimal requirements on location. They can provide very detailed information about the air situation even in situations where radar coverage cannot effectively be achieved. This talk presents some of technical aspects of ADS-Band discusses how to build a reliable ATC infrastructure with ADS-B.

This colloquium is presented jointly by the Department of Computer Science and the Department of Electrical and Computer Engineering.

Tuesday, May 15, 2007, 3:30pm, Ungar 402 -- Coffee at 3:00

Elisha Sacks
Department of Computer Science
Purdue University

will present

Computer-Aided Mechanical Design Using Configuration Spaces

I will describe our research in computer-aided mechanical design. Mechanical design is the task of devising an assembly of parts that performs a function reliably and economically. It is a ubiquitous activity with applications in mechanical, electrical, and biomedical engineering. Our research addresses the core design task of kinematic analysis. Kinematic analysis determines the positions and orientations at which the parts of a system touch and the ways that the touching parts interact. It is a computational bottleneck in mechanical design, especially in systems with complex part shapes, tight fits, and contact changes. We have developed a general kinematic analysis method based on configuration space computation. The method incorporates efficient algorithms for function validation, tolerancing, parameter synthesis, and redesign. I will illustrate these capabilities on industrial applications in gearshift design (with Ford Motors) and micro-mechanism design (with Sandia National Laboratory).

Joint work with Leo Joskowicz, Hebrew University.

Wednesday, April 25th 2007, 5:00pm, Ungar 411

Ronda Edwards
Department of Computer Science
University of Miami

will give a project presentation

A Digital Scrapbook

In my presentation I will be discussing the digital scrapbook that I created for the Filipino Student Association. I will be giving some background information about the project and why I chose this project, and I will show a brief demonstration of the product. I will then discuss how I made the book using Photo Impact Pro and Macromedia Flash and give an overview of the creation process. After that, I will discuss the additions to the project and show demonstrations of their functionality. This project will be placed on the website of the Filipino Student Association so that all members can enjoy the scrapbook and the game. It will also be accessible to non-members who can learn about the club and about Filipino culture.

Gonzalo Martinez
Department of Computer Science
University of Miami

will give a project presentation

Rating TPTP Problems and Systems

My project consists on creating a system for rating TPTP problems and SPC systems. The program is provided a file, containing a huge amount of data about the systems and the problems. This data contains, for example, for each system the problems it has been tested with and if the system can solve the problem or not. With this data, my program has to rate the systems and the problems. The problems are given a rating depending on the difficulty, and the systems are given a rating depending on the number of problems they solve. Once I have calculated the ratings, I gave to print it with a specific format. The issue of this project is the efficiency. Because of the large amount of data, the system has to be very efficient in order to process all the data and give the output in a reasonable time.

Jorge Jauregui
Department of Computer Science
University of Miami

will give a project presentation

Techniques and Issues to Consider When Developing a Flexible Game Engine

MiamiVideo games have been around for decades and have become a staple of entertainment across the globe. Ever wonder how they work? It all begins with the game engine. This is a crucial component of any serious game and is responsible for everything from graphical output to level management. This talk is aimed at anyone with a desire to see beyond what the player sees and learn techniques to use when building their own game engine. Topics include game state management, data organization, sprite animation, the 3D rendering pipline, character mechanics, input and sound as well as game implementation techniques. The discussion will conclude with a demo of an action adventure RPG named "Hidden in Time" developed using Jorge Jauregui's Substance Game Engine.

This is another in the Departmental Pizza Seminar series.

Wednesday, April 18th 2007, 5:00pm, Ungar 411

Beau Silver
Department of Computer Science
University of Miami

will give a project presentation

Preparing to Program in the Audio Industry

The wide spread use of computers and digital technology has lead way to the binding of audio and computer programming. Computer software now runs audio recording (Pro Tools), audio storage (hard drives) and audio reproduction (Ipod). Traditional computer programming teaches the fundamentals of manipulating numbers, printing out text and maybe even reading from and writing to a file. This lays the foundation for a student to perform audio programming, but how does one go to the next step? How is it possible to read and write from audio files? How are digital audio files encoded? How is it possible to synthesize audio on computer's sound card? The answer to these questions can be found by using C libraries to manipulate audio and by preparing to program complex software by learning the benefits of C++.

Frank Diaz
Department of Computer Science
University of Miami

will give a project presentation

TBA

This is another in the Departmental Pizza Seminar series.

Wednesday, April 11th 2007, 5:00pm, Ungar 411

Geoff Sutcliffe
Department of Computer Science
University of Miami

will present

How to Give a Successful Talk

Almost everyone has to give a public presentation at some time. Graduate students have to defend their theses, research students have to present their results, and many jobs require presentations. It is common to have little or no experience when you give your first presentation, and you may even be a little nervous! This talk is aimed at (graduate, research, and other) students. It describes how to give a successful talk, in a simple standard format. This talk does not try to teach general speaking skills, nor impose any personally preferred techniques. It covers the structure of a talk, the use of visual aids, speaking technique, and how to cope with questions.

This is another in the Departmental Pizza Seminar series.

Wednesday, April 11th 2007, 5:00pm, Ungar 411

Wednesday, April 4th 2007, 5:00pm, Ungar 411

Geoff Sutcliffe
Department of Computer Science
University of Miami

will present

SRASS - a Semantic Relevance Axiom Selection System

This talk describes the design, implementation, and testing of a system for selecting necessary axioms from a large set also containing superfluous axioms, to obtain a proof of a conjecture. The selection is determined by semantics of the axioms and conjecture, ordered heuristically by a syntactic relevance measure. The system is able to solve many problems that cannot be solved alone by the underlying conventional automated reasoning system.

This is another in the Departmental Pizza Seminar series.

Monday April 2nd 2007, 5:00pm, Ungar 402

Mehmet Koyuturk
Department of Computer Science
Purdue University

will present

Comparative Analysis of Molecular Interaction Networks

Emergence of high-throughput experiments and resulting databases capture relationships and interactions between biomolecules. These interactions enable modeling and analysis of a cell from a systems perspective - generally using network models. In this talk, we focus on development of computational tools and statistical models for comparative analysis of molecular interaction networks. These tools target understanding of functional modularity in the cell by extracting novel information from massive amounts of interaction data, through integration of cellular organization, functional hierarchy, and evolutionary conservation.

We first discuss the problem of identifying conserved sub-networks in a collection of interaction networks belonging to diverse species. The main algorithmic challenges here stem from the NP-hard subgraph isomorphism problem that underlies frequent subgraph discovery. Three decades of research into theoretical aspects of this problem has highlighted the futility of syntactic approaches, thus motivating use of semantic information. Using a biologically motivated homolog contraction technique for relating proteins across species, we render this problem tractable. We experimentally show that the proposed method can be used as a pruning heuristic that accelerates existing techniques significantly, as well as a standalone tool that conveys significant biological insights at near-interactive rates.

With a view to understanding the conservation and divergence of modular substructures, we also develop network alignment techniques, grounded in theoretical models of network evolution. Through graph-theoretic modeling of evolutionary events in terms of matches, mismatches, and duplications, we reduce the alignment problem to a graph optimization problem and develop heuristics to solve this problem efficiently. In order to assess the statistical significance of the patterns identified by our algorithms, we probabilistically analyze the distribution of highly connected and conserved subgraphs in random graphs. Our methods and algorithms are implemented on various platforms and tested extensively on a comprehensive collection of molecular interaction data, illustrating their effectiveness in terms of providing novel biological insights as well as computational efficiency.

Mehmet Koyuturk received his B.S. (1998) and M.S. (2000) degrees in Electrical and Electronics Engineering, and Computer Engineering, respectively, from Bilkent University, Turkey. During his graduate studies at the Department of Computer Science at Purdue University, he worked on a number of problems in the areas of Computational Biology and Bioinformatics, Parallel and Distributed Computing, and Scientific Computing. His thesis focused on algorithmic and analytical aspects of comparative analysis of biological networks. His collaborations with domain experts in this area resulted in several significant publications and software tools. Since receiving his Ph.D. in August 2006, he has been a post-doctoral research associate in the same department.

This is joint work with Yohan Kim, Shankar Subramaniam (University of California, San Diego), Wojciech Szpankowski, and Ananth Grama (Purdue University) and is supported by the National Institutes of Health.

Wednesday March 21st 2007, 5:00pm, Ungar 402

Jianhua Ruan
Department of Computer Science and Engineering
Washington University in St. Louis

will present

Community Discovery and Analysis in Biological Networks

Many complex systems can be represented as networks, where the nodes are the elements in a system, and the edges represent relationships between pairs of elements. Examples include social networks, protein-protein interaction networks, and the world-wide web. Much effort has been devoted to the study of topological properties that seem to be common to many networks, such as the small-world property and power-law degree distributions. Another important property that has drawn a great deal of interest recently is the so-called community structure, i.e. the existence of some natural division of a network such that nodes in each sub-network are highly associated among themselves, while having relatively fewer/weaker connections with the rest of the network. Automatically identifying such communities is fundamental for revealing the relationships between the structure and function of complex networks, with many applications in different disciplines such as sociology, biology, and computer science.

In this talk, I will discuss the conceptual and algorithmic challenges in community discovery, i.e. how to define communities, and how to identify them, and present our recent contributions in addressing these challenges. I will show several real applications of our methods in biology and other areas, and demonstrate the advantages of our methods against the existing approaches. Furthermore, I will show that the members of the same community often have very strong functional ties, which shed lights on the organizing principles of the system, and provide key insights about the functions of some previously uncharacterized members. I will conclude with an overview of my research interests and plan of future works.

Bio: Jianhua Ruan is currently a PhD candidate in the Department of Computer Science and Engineering at Washington University in St Louis. He received his B.S. degree (1998) in Biology from University of Science and Technology of China, and M.S. degree (2002) in Computer Science from California State University. His current research interests in computational biology and bioinformatics include (1) structural and functional properties of biological networks, (2) reverse-engineering of gene transcriptional regulatory networks, and (3) RNA structures and functions.

Monday March 19th 2007, 5:00pm, Ungar 402

Dimitris Papamichail
Computer Science Department
State University of New York at Stony Brook

will present

Classification and Synthesis of Genomic Sequences

The talk consists of two main parts:

Genomic sequence classification and analysis
Microorganisms dominate the biosphere, yet most have not been identified or studied. Traditional methods for culturing and characterizing microorganisms limit analysis to those that will grow under laboratory conditions, which represent less than 1% of all microorganisms. In this presentation, I will show an oligonucleotide (k-mer) classification method based on conditional probabilities, which performs substantially better than other known methods. I will demonstrate that k-mer distributions are well-preserved among related strains/species and that it is possible to identify bacterial species, even from mixed populations, via k-mer distributions, using modest amounts of sample sequence. Population analysis is persistently challenging but important to microbiologists, leading to determination of diversity and function of members of microbial communities. With our bioinformatics methods we can achieve this analysis via a homology based method for robust phylotype determination, enhancing closely related sequence associations and a methodology for achieving more accurate richness estimation, using different clustering criteria.

Genomic sequence design and synthesis
The emerging field of synthetic biology is broadly defined as the area of intersection of biology and engineering that focuses on the modification or creation of novel biological systems that do not have a counterpart in nature. We explored the problem of designing the provably shortest genomic sequence to encode a given set of genes by exploiting alternate reading frames. Here I will present an algorithm for designing the shortest DNA sequence simultaneously encoding two given amino acid sequences. We have shown that the coding sequences of naturally occurring pairs of overlapping genes approach maximum compression, as well as investigated the impact of alternate coding matrices on overlapping sequence design. Working with the group that achieved the first genome-level synthesis of a virus, we have designed, synthesized, and evaluated two new variants of poliovirus to serve as vaccines. Specifically, we seeked weakened but viable strains that may be used for preparations of a killed poliovirus vaccine. Our designs result in a virus with roughly 100-fold lower specific infectivity than the wildtype virus. Here I will present the theory behind gene design in the context of optimizing a DNA sequence for particular desired properties while simultaneously coding for a given amino acid sequence.

Wednesday, March 7th 2007, 5:00pm, Ungar 411

Dimitri Boelaert-Roche
Department of Computer Science
University of Miami

will present

Exploring Modern Game Graphics

This overview of current graphic algorithms in the Game Industry starts with an introduction to video card capabilities highlighting the differences between software rendering, the old Fixed Function Pipeline, and the new standard, the Programmable Pipeline. Graphical approaches to game aesthetics such as Shadow mapping, Real-time lighting, Day / Night cycles, Terrain Rendering, and Cel-Shading will be demonstrated using "shader programs" on graphics hardware, utilizing the programmable pipeline that is prevalent in today's consumer desktop and video game consoles.

This is another in the Departmental Pizza Seminar series.

Wednesday February 21st 2007, 5:00pm, Ungar 402

Professor Mitsunori Ogihara
Department of Computer Science
University of Rochester

will present

Data Mining for Exploration of Databases

Data mining is a new area of computer science that integrates ideas from other areas of computer science, such as algorithms, complexity theory, computer systems, databases, machine learning, and network systems. The goal of data mining is to discover interesting information in large databases. In this talk I will present some of my past and current research work in the area of data mining. I will first introduce association mining and sequence mining, both of which have been known to be quite useful in businesses. In the former the goal is to enumerate all frequent combinations of data attributes appearing in a database, in the latter the data is time- stamped and the goal is to find all frequent sequences of frequent combinations of data attributes. I will present how these data mining tasks can be used for slightly different purposes, for finding similarities among database and for predicting future events. I will then switch a gear and talk about some work on gene expression data analysis. Semi-supervised learning refers to machine learning techniques for building more accurate models creatively using unlabeled data. I will present how semi-supervised learning can be used to improve the accuracy of gene function classification. I will then speak about multi-class sample classification using gene expression data.

Wednesday September 27th 2006, 3:30pm, Ungar 402

Dr. Russ Miller
Hauptman-Woodward Medical Research Institute
State University of New York at Buffalo

will present

Cyberinfrastructure and Molecular Structure Determination

In this talk, we give an overview of Buffalo's Center for Computational Research (CCR), one of the leading academic supercomputing sites in the world. CCR supports work in the physical sciences, engineering, life sciences, and visualization. We will also present an overview of our design and implementation of the ACDC-Grid, an extensive, multi-institutional proof-of-concept grid that is being expanded to incorporate leading academic and non-profit organizations in New York State. We will discuss our efforts in grid monitoring, predictive scheduling, grid-enabling application templates, backfill detection and optimization, data repositories and operations, dynamic and automated allocation of resources, and our dynamic firewall, to name a few. We will then present an overview of our Shake-and-Bake method of molecular structure determination and discuss critical parameters that have been optimized in our SnB instantiation of this procedure. Finally, we will discuss the functionality of the ACDC-Grid that enables a cost-effective and transparent grid-based port of SnB, including grid-enabled optimization via data repositories, data mining, and intelligent generation of jobs to be run by the daemon overseeing the parameter optimization routine.

Dr. Miller is UB Distinguished Professor of Computer Science and Engineering at SUNY-Buffalo and Senior Research Scientist at the Hauptman-Woodward Medical Research Institute. Shake-and-Bake was listed on the IEEE poster "Top 10 Algorithms of the 20th Century". The work presented in this talk was supported by NSF, NIH, DOE, NYS, the Oishei Foundation, the Wendt Foundation, NIMA, Dell, IBM, and HP. This talk represents joint work with many students, staff, and colleagues, who are listed in the talk.

Wednesday, September 20th 2006, 5:00pm, Ungar 411

Dr Josef Urban
Department of Theoretical Computer Science and Mathematical Logic
Charles University, Prague, Czech Republic

will present

A Short Introduction to Mizar for Mathematicians and Computer Scientists

You'll hear a short summary of the current state of the art in formal and computer-checked mathematics, as well as the motivation and history of formalization efforts. A lightweight overview of one of the leading formalization systems and projects - Mizar - will follow. Finally I'll talk about Mizar and its large formal knowledge base as a very interesting object for various Artificial Intelligence methods, and their potential combinations.

This is another in the Departmental Pizza Seminar series.

Wednesday, September 13th 2006, 5:00pm, Ungar 411

Weina Shen
Department of Computer Science
University of Miami

will present

Automated Proofs of Equivalence of Modal Logic Systems

The equivalence of different axiomatizations of modal logics is well known and documented in the literature. e.g., S5 may be equivalently axiomatized by PC + necessitation + K (distribution) + M + 5, and by S10 + M6 + S3 + M9 + B. In order to automatically prove the equivalence of two modal systems using first-order theorem provers, we provide a series of encoding rules to represent modal logic syntax in first-order syntax, and a definitional approach to encode each proof problem into a TPTP file for mechanical deduction in a first-order theorem prover.

This is another in the Departmental Pizza Seminar series.

Monday, July 3rd 2006, 2:00pm, Ungar 506

Dr. Gabriel Tsechpenakis
Center for Computational Biomedicine, Imaging & Modeling
Rutgers University

will present

Computer Vision: From the Macro- to the Micro-world

Computer Vision is the study and application of methods that allow computers to "understand" image and video content, or content of multidimensional data. The term "understand" means that specific information is being extracted from the image or video data for a specific purpose: either for presenting it to a human operator (e. g., if cancerous cells have been detected in a microscopy image), or for controlling a process (e. g., an autonomous vehicle). Computer Vision can also be described as the complement of biological vision. In contrast to biological vision and visual perception systems though, Computer Vision enables for quantitative analysis and representation of data, and therefore it is becoming more and more essential to a wide variety of research fields.

In my talk, I will focus on different methods I have been developing in the last eight years for different applications, varying from Human-Computer and Human-Human Interaction (HCI and HHI) to 3D shape reconstruction of human organs. Specifically, the first part of my talk will be on 2D and 3D tracking of rigid, deformable and articulated objects in monocular video sequences, and how low-level features (e.g., grayscale image intensity) lead to motion and shape extraction, video segmentation, and action/event recognition. I will present the various applications of my methods, from object-based video coding to American Sign Language recognition. In the second part of my talk I will present how Computer Vision methods are applied to biomedicine. I will describe the theoretical framework of LocoWorm, a software I have developed and is currently in use for multiple tracking, locomotion features extraction and classification for aging and phenotypical analysis of C. elegans nematodes. I will also present the problem of 3D reconstruction of alveoli from lung tissue scans, and I will give the general framework of my approach. Finally, I will describe a novel method for efficient medical image segmentation, which uses local and global shape and texture information. This method can be seen as a fusion between region growing and active shape models (ASMs).

Wednesday, April 26th 2006, 5:00pm, Ungar 411

Luis Portillo and Antonio Ortells
Department of Computer Science
University of Miami

will present

Design and Implementation of a 3D Video Game

This project is about designing and implementing a complete video game. By doing this we discovered all the ins and outs of making a complete software product. The game is a 3D first person adventure in which you will have to solve several puzzles and fight versus fully armed enemies; it is developed in C++ with Open GL for the graphics.

... and ...

Joshua Chudnovsky
Department of Computer Science
University of Miami

will present

Using Visual Basic Based GUI for Design of Code Conversion

The main purpose of the development of the application was to create a tool that can be used in the Medical Offices as a help to billing or coding personnel. Program offers selection screens to select options needed for the transfer and possible conversion of ICD-9 codes (International Classification of Diseases, 9th Revision). Problems related to this work include: maintaining the integrity of the data structure used to manipulate code, necessity of changes required by the need for conversion. There were some programming and data management issues, that will be discussed during the presentation.

... and ...

Richard Cayemitte, Tom Robinson and Gabriel Rosauro
Department of Computer Science
University of Miami

will present

The Necessity for Hats: A HeadWARe Post-Mortem

What happens when you put three code ninjas to find a solution to world hunger? They make a game of course! HeadWARe is a 3D strategy game where squads of robots wearing nifty hats battle out for world domination. Coded from the ground up during a period of two semesters, by four able hands, HeadWARe was conceived and created to showcase the talent and work that drive the people who created it. During this presentation we will walk you through the various stages of development and discuss the logistic and design challenges encountered by the team during the execution of this project.

This is another in the Departmental Pizza Seminar series.

Wednesday, April 19th 2006, 5:00pm, Ungar 411

Richard Cayemitte, Ben Li
Department of Computer Science
University of Miami

will present

A Software Tool to Visualize Placement, Compaction and
Configuration Space Algorithms on an Arrangement of Polygons

This aim of the project was to extend an existing software tool to visualize the results of placement, compaction, as well as configuration space algorithms that were written by Dr Victor Milenkovic. Placement and compaction algorithms are heuristics to minimize the total amount of space occupied by a set of polygons. These heuristics are run with certain restrictions placed on each polygon. For a configuration space, The configuration of an unfixed 2-D polygon relative to one or more 2-D polygons fixed on a canvas can be expressed as a triple, (x,y,θ), with (x,y) specifying the polygon's coordinates and θ specifying its rotation. The configuration space of the same polygon relative to the other polygons can be expressed as the set of all configurations, or triples, (x,y,θ), that the unfixed polygon can take. This configuration space can be decomposed into a blocked space - the set of triples (x,y,θ) in which some polygons overlap, and a free space, its complement. Professor Milenkovic implemented algorithms to calculate this decomposition, and we added an interface to an existing Java application to visualize this boundary, in 3-D (i.e., the set of (x,y,θ) ), with a 2D cross-section (i.e., only the set of (x,y) ). The interface also allows the user to control the translation and rotation of the unfixed polygon and see its current configuration as a point in the 3-D space and 2-D cross-section. Vice versa, the user can control the configuration point and immediately see the actual, updated arrangement of the polygons.

... and ...

Juan Panadero Ruiz
Department of Computer Science
University of Miami

will present

Programming Cell Phone Web Interface using Flash

The presentation will include the following topics:

• basic idea for the project involving web based interface.
• technology involved.
• programming technique invovled.
• limitations of the cell phone screen size.
• challenges in making decision which software product to use for the programming.

... and ...

Mario Russo
Department of Computer Science
University of Miami

will present

Backtesting Financial Trading Systems

A trading system is simply a group of specific rules, or parameters, that determine entry and exit points for a given equity. It seems that everywhere you look, you see advertisements for software promising accurate buy and sell signals and profits with every trade - all with minimal time and effort. But, what if you took the time to select some predefined financial formulas and created your very own trading system, can your trading systems offer profitable methods of trading? Well now you can check years and years of past performance of the system with my Backtesting software.

This is another in the Departmental Pizza Seminar series.

Monday, April 17th 2006, 5:00pm, Ungar 402

Orhan Camoglu
Department of Computer Science
University of California, Santa Barbara

will present

Discovering Functional Relationships among Proteins Using Computational Techniques

Recent advances in molecular biology have resulted in immense amounts of biological data (DNA sequences, protein sequences and structures, gene expression data, and protein interaction data). As a result of this growth, scalable techniques to mine and analyze this data have become paramount. These new techniques employ principles in many different areas of computer science, especially databases, data mining, machine learning and graph theory. They also impose new challenges. This talk focuses on applying database principles and computational techniques on biological data to infer functional properties of proteins. Functional and evolutionary relationships between proteins can be discovered via structure comparison. As the sizes of experimentally determined and theoretically estimated protein structure databases grow, there is a need for scalable search techniques. In this talk I will present a technique that extracts feature vectors on triplets of SSEs (Secondary Structure Elements) of proteins and uses an index structure to answer similarity queries. Currently, a wide range of information sources is available to provide evidence about the functional relationships among proteins. I will present an automated functional classification methodology that uses these information sources. Protein interaction networks provide a different view of protein functional relationships based on their interaction properties and can be used to infer intricate relations, such as pathway and complex membership. I will present analysis techniques for mining interaction networks to capture functional relationships among proteins.

Wednesday, April 12th 2006, 5:00pm, Ungar 411

Kevin Moynihan
Department of Computer Science
University of Miami

will present

Developing Aesthetic Computer Generated Drawings through Artificial Evolution

This talk discusses the production of visually appealing computer-generated drawings, through the emulation of human drawing techniques and artificial evolution. Employing a population of line drawn objects from user-input, the Darwinian process of evolution produces original drawings with increasing levels of user satisfaction.

... and ...

Yury Puzis
Department of Computer Science
University of Miami

will present

Automated Generation of Interesting Theorems

In the logical theory of a set of axioms there are many boring logical consequences, and scattered among them there are a few interesting ones. The few interesting ones include those that are singled out as {\em theorems} by experts in the domain. This paper describes the techniques, implementation, and results of an automated system that generates logical consequences of a set of axioms, and uses filters and ranking to identify interesting theorems among the logical consequences.

This is another in the Departmental Pizza Seminar series.

Wednesday, April 5th 2006, 5:00pm, Ungar 411

Dr Raju Rangaswami
Department of Computer Science
Florida International University

will present

Operating System Security: Host-based Intrusion Detection

An intrusion occurs as a sequence of operations that may individually affect various aspects of a computer system. Current host-based intrusion detection systems that monitor the operations within a single subsystem (e.g. file system or network) suffer from myopic views of system state. Such local views can then lead to incorrect inferences about system status. In this talk, I will present a holistic system-centric approach to intrusion detection, which monitors the global state of the host. I will present the challenges that face building such a system and present the salient features of our approach that allow overcoming these challenges.

Dr. Raju Rangaswami is an Assistant Professor in the School of Computing and Information Sciences at Florida International University. He received M.S. and Ph.D. degrees from the University of California at Santa Barbara and his B.S. degree from the Indian Institute of Technology, Kharagpur, India. He conducts research in operating systems, security, storage systems, and real-time systems. His research is supported by the National Science Foundation. More details can be found at http://www.cs.fiu.edu/~raju/

This is another in the Departmental Pizza Seminar series.

Wednesday, March 29th 2006, 5:00pm, Ungar 411

Dr Xiaodong Cai
Department of Electrical and Computer Engineering
University of Miami

will present

Stochastic Modeling and Simulation of Gene Networks

Gene networks are traditionally modeled as a deterministic dynamic system using ordinary differential equations. However, it was found recently that gene expression is best viewed as a stochastic process, since gene expression consists of a series of events that involve a small number of molecules such as DNA, RNA, and proteins. Therefore, any computational modeling and simulation of a gene network should take into account such stochasticity in gene expression. In this talk, I will first introduce stochastic modeling of gene networks from both biological and computational perspective, and then briefly describe characterization of the stochastic dynamics of gene networks or more general any coupled chemical reaction systems. I will also briefly review stochastic simulation algorithms (SSA) including the exact SSA, approximate but more efficient leaping methods, and multiscale SSAs. Finally, I will present an efficient SSA called K-leap SSA, that we developed recently, and compare its performance with that of other SSAs.

This is another in the Departmental Pizza Seminar series.

Wednesday, March 22nd 2006, 5:00pm, Ungar 411

Dr Tao Li
Department of Computer Science
Florida International University

will present

Mining Log Data for Computing System Management

Over the years, the advancements in science and technology have led to the increased complexity in computing systems. The systems are thus becoming increasingly more complex with growing number of heterogeneous software and hardware components, increasingly more difficult to monitor, manage and maintain. There is thus a pressing need for automatic and efficient approaches to monitor and manage complex systems. In this talk, I will first describe our research effort on mining log data for automatic system management. In particular, I will talk about two research components of the holistic effort: (a) novel text mining approaches to transform system log data in disparate formats and contents into a canonical form; (b) new temporal data mining techniques to discover relationships among event data. I will also discuss some research directions and future work on this project. If time permits, the talk will also give an overview of some other research projects that we are currently working on such as microarray data analysis, matrix factorization for text clustering, music information retrieval, adaptive workflow management.

Dr. Tao Li is currently an assistant professor in the School of Computer Science, Florida International University. He received his Ph.D. in computer science from the Department of Computer Science, University of Rochester in July 2004. His research interests are in data mining, machine learning, information retrieval, and bioinformatics. He is a recipient of NSF Career Award, IBM Faculty Award, and IBM SUR (Shared University Research) Award. More information about him can be found at http://www.cs.fiu.edu/~taoli.

This is another in the Departmental Pizza Seminar series.

Wednesday, February 22nd 2006, 5:00pm, Ungar 402

Luke Huan
Department of Computer Science
University of North Carolina

will present

Discovering Patterns in Families of Protein Structures

The rapidly growing body of 3D protein structures provides new opportunities to study the link between protein structure and protein function. Local structural comparison, which aims to identify arrangements of amino acids common to a family of proteins, is a way to discover structural features linking protein structures to their function. Traditional approaches to local structural comparison of proteins operate in a pair-wise fashion and have prohibitive computational cost when scaled to families of proteins. This talk describes my research on graph-based representations of protein structure and new subgraph mining algorithms to identify recurring structural patterns shared by many members of a family of proteins. A statistical measure is defined to evaluate the significance of each pattern's association with a family of proteins compared to all known protein structures.

Two collaborations with domain experts at the UNC School of Pharmacy and the UNC Medical School illustrate the use of these techniques. The first is to predict the function of several newly characterized protein structures based on the amino acid arrangements they contain. The second is to identify conserved structural features in evolutionarily related proteins. Such structures may play an important role in folding. This talk presents results from both applications and concludes with my future research plans in the areas of data mining, proteomics, and systems biology.

Mr. Huan is a Ph.D. candidate in the Department of Computer Science and a member of the Bioinformatics and Computational Biology Training Program at the University of North Carolina. He received his B.S. in Biochemistry from the Peking University, China in 1997 and an M.S. in Computer Science from the Oklahoma State University in 2000. He worked at Argonne National Laboratory and Nortel before he joined UNC and has current collaborations with GlaxoSmithKline. His research interests include data mining, proteomics, systems biology, and high performance computing. He was a recipient of the UNC Scholar of Tomorrow Fellowship in 2001 and the Alumni Fellowship in 2005.

Wednesday, February 15th 2006, 5:00pm, Ungar 411

Melanie Rieback
Department of Computer Science
Vrije Universiteit, Amsterdam

will present

Radio Frequency Identification: Applications, Threats, and Countermeasures

Radio Frequency Identification (RFID) is a popular contactless identification technology which has been hyped as the "next generation barcode". These tiny remotely-powered computer chips have already been integrated into consumer goods, passports, public transportation tickets, and even people. Because most RFID tags lack privacy enhancing technologies or cryptography, malicious parties can use RFID technology for nefarious ends like theft, stalking, and behavioral profiling.

This presentation will discuss RFID technology, the security and privacy threats which it faces, and proposed countermeasures. I will highlight one particular new solution, the RFID Guardian, that is the focus of a collaboration between researchers at the Vrije Universiteit and the Technical University of Delft.

This is another in the Departmental Pizza Seminar series.

Wednesday, February 15th 2006, 2:30pm, Ungar 411

Dr My (Tracy) Thai
Department of Computer Science and Engineering
University of Minnesota

will present

Energy Efficient Coverage in Wireless Sensor Networks

Wireless Sensor Networks (WSNs) are believed to be one of the emerging technologies that would change the world. WSNs can be used in a wide range of potential applications in both military and civil. Since sensor nodes are battery powered, energy efficiency is of primary concern in WSNs. To prolong a sensor network lifetime, energy efficiency must be considered in almost every aspect of sensor network design. In this talk, I will address the maximum lifetime coverage problem, which is to maximize network lifetime while a sensor network performs the coverage tasks. I will present a novel mathematical model and a Linear Programming based algorithm to approximate the solution of this problem. I will also acknowledge the relation between this problem and the domatic number problem.

In addition, toward the end of this talk, I will highlight some important results of my other work in both wireless networks and computational biology.

Wednesday, February 8th 2006, 5:00pm, Ungar 411

Christian Duncan
Department of Computer Science
University of Miami

will present

The Visualization of Information

The drawing of information, often represented as a graph, onto a medium such as a computer screen is a fundamental yet challenging problem in computer science. The algorithms developed incorporate techniques from computational geometry, graph theory, optimization, and computer graphics.

In general, the underlying problem is to take a graph composed of vertices and edges between the vertices and place (embed) these vertices and edges on the screen in a visually appealing and informative manner. The precise definition of "appealing" and "informative" are of great debate in the graph drawing community, but certain standards are almost universally accepted. For example, graphs with multiple edge crossings are considered less appealing and more confusing than similar graphs without any crossings.

In this talk, we shall introduce the area of graph drawing, give a brief description of basic results and problems in the field, and discuss some of our own recent results particularly in the area of simultaneous graph embeddings.

This is another in the Departmental Pizza Seminar series.

Wednesday, January 18th 2006, 4:30pm, Ungar 411

Dr Sawsan Khuri
The Dr. John T. Macdonald Foundation Center for Medical Genetics
Miller School of Medicine
University of Miami

will present

Bioinformatics as Relevant to Medical Genetics

Bioinformatics has become a buzz word to mean almost anything and everything to do with the interface of computers and biology. From the perspective of the discipline of Medical Genetics, there are two clear applications for this interface. One is in the database arena, linking a patient's genetic and family history information with data on the disease or syndrome that this patient presents with. The second is in understanding how a genetic problem actually translates to a disease. In other words, analysis of the gene sequence, of the protein that this gene makes, and of how this gene is switched on and off. These methods fall mostly under the title of "genomics", and have had a meteoric rise since the completion of the human genome sequence. This talk will summarize the advances made in bioinformatics for Medical Genetics, and highlight the areas where there is room for improvement.

This is another in the Departmental Pizza Seminar series.

Wednesday, December 14th 2005, 5:00pm, Ungar 402

Dean Felch
Department of Computer Science
University of Miami

will present

Online Medical Record Security

Background: Ever since patient medical information has been put into computerized databases, there has been a plethora of security issues related to information privacy. Now, with the HIPAA act, medical facilities are being held responsible for the security of protected information. While many of the cryptographic security issues have been solved, there is still vulnerability from unauthorized users who have obtained access to an online medical record database using legitimate login information.

Abstract of Solution: To solve this problem I have designed a modification specification to database source code which requires certain "tags" to be added to queries. An unauthorized user who has gained access to the database remotely through someone else's authorized login information would receive fictitious query results when he/she sends "untagged" queries to the database. Furthermore, the sensitive information within the patient medical record would be stored in a separate table from the patient demographic information such that an unauthorized user would have to perform a table join within the query.

This is another in the Departmental Pizza Seminar series.

Thursday, December 8th 2005, 5:00pm, Ungar 402

CSC531 - Software Engineering Class
Department of Computer Science
University of Miami

will present

The Software Engineering of an
Electronic Medical Record System

We will explain the process of analyzing, planning, implementing and testing an Electronic Medical Record (EMR) system from a software engineering perspective. An EMR system is responsible for maintaining patient medical records as well as facilitating other office procedures (e.g. scheduling, billing). Part of the systems engineering process requires communication between team members to revise code and documentation. To streamline this process we developed a project management website that allows team members to upload their code and documentation remotely for review by the other team members.

This is another in the Departmental Pizza Seminar series.

Wednesday, November 9th 2005, 5:00pm, Ungar 506

Dr Mohamed Iskandarani
Rosenstiel School of Marine and Atmospheric Science
University of Miami

will present

Ocean Modeling with High Order Finite Element Methods:
Challenges and Opportunities

Oceanic simulations have traditionally relied on structured-grid, finite difference methods to discretize the hydrostatic primitive equations governing large scale ocean flows. The complex coastlines of ocean basins, and the geometrically flexibility inherent in unstructured grids has spurred ocean modelers to pursue the development of finite element based models. Here we sketch the outlines of one such code: the spectral element ocean model (SEOM). The spectral element methods combine the geometrical flexibility of low-order finite element methods and spectral methods. Its advantages include high accuracy, very-low numerical dissipation, dense computational kernels at the element level and sparse neighbor-to-neighbor communication; the latter combination leads to extremely good scalability on parallel computers. In the talk the basis of the method will be presented, and sample two-dimensional simulations will be shown. The final part of the talk will revolve around the challenges of three-dimensional ocean modeling.

This is another in the Departmental Pizza Seminar series.

Wednesday, September 14th 2005, 5:00pm, Ungar 411

Steven Trac
Department of Computer Science
University of Miami

will present

Robust Topologically Invariant Set Operations on
2D Semi-Algebraic Sets

We present robust floating point algorithms for performing topologically invariant set operations on 2D semi-algebraic sets. A 2D semi-algebraic set is a set of cells from an arrangement of segments of algebraic curves. These algorithms transform and overlay multiple arrangements invariantly. Because the algorithms are implemented in floating point, certain topological changes are unavoidable. For instance, after a floating point Euclidean transformation, two very close but distinct points may transform into the same point. The topological changes that can result from floating point rounding are analyzed. The algorithms are invariant except for these types of changes. We discovered that the unavoidable topological changes in our robust topologically invariant algorithms occur rarely in testing. We ran tests on our topologically invariant algorithms against the original algorithms by Milenkovic and Sacks, comparing computing time and accuracy.

This is another in the Departmental Pizza Seminar series.

Friday, April 8th 2005, 2:30pm, Ungar 402

Professor Toniann Pitassi
Department of Computer Science
University of Toronto

will present

Learnability and Automatizability

In this talk we prove new upper and lower bounds on the proper PAC learnability of decision trees, DNF formulas, and intersections of halfspaces. Several of our results were obtained by exploring a new connection between automatizability in proof complexity and learnability. After explaining this basic connection, we will prove the following new results.

1. We give new upper bounds for proper PAC learning of decision trees and DNF, based on similar known algorithms for automatizability of Resolution.
2. We show that it is not possible to PAC learn DNF by DNF in polynomial-time unless NP ⊆ BPP. We also prove the same negative result for proper PAC learning of intersections of halfspaces.
3. We show that decision trees cannot be proper PAC learned, under the assumption that SAT is not solvable by subexponential-size circuits.

This is joint work with Misha Alekhnovich, Mark Braverman, Vitaly Feldman, and Adam Klivans.

Wednesday, March 30th 2005, 5:00pm, Ungar 411

Geoff Sutcliffe
Department of Computer Science
University of Miami

will present

Discrete Event Calculus Deduction using
First-Order Automated Theorem Proving

The event calculus is a powerful and highly usable formalism for reasoning about action and change. The discrete event calculus limits time to integers. This paper shows how discrete event calculus problems can be encoded in first-order logic, and solved using a first-order logic automated theorem proving system. The following techniques are discussed: reification is used to convert event and fluent atoms into first-order terms, uniqueness-of-names axioms are generated to ensure uniqueness of event and fluent terms, predicate completion is used to convert second-order circumscriptions into first-order formulae, and a limited first-order axiomatization of integer arithmetic is developed. The performance of first-order automated theorem proving is compared to that of satisfiability solving.

This is joint work with Erik Mueller of IBM Thomas J. Watson Research Center.

This is another in the Departmental Pizza Seminar series.

Wednesday, March 23rd 2005, 5:00pm, Ungar 411

Yuan Zhang
Department of Computer Science
University of Miami

will present

The Use of Lemmas for Solving
Hard Automated Theorem Proving Problems

Using lemmas has been proved to be an effective approach for assisting ATP systems to solve hard problems. Useful lemmas can provide valuable guidance in the proof search, and help construct the proof by filling in intermediate steps. However, the formulae supplied to an ATP system as lemmas are not all necessarily useful. Unuseful lemmas act as noise, disturbing the search for the proof. It is therefore necessary to develop lemma management techniques that identify useful lemmas, and help an ATP system to use the useful lemmas to its advantage. We have designed three lemma management techniques, implemented them, and illustrated their potential with example problems. It has been shown that, with these three lemma management techniques, the problem-solving ability of an ATP system can be improved.

This is another in the Departmental Pizza Seminar series.

Wednesday, March 9th 2005, 5:00pm, Ungar 411

Dr. Ziya Arnavut
SUNY, Fredonia

will present

Inversion Coding

The Block Sorting Coder (BSC) described by Burrows and Wheeler has received considerable attention. BSC achieves a compression ratio closer to PPM, but with a faster execution speed than PPM.

An essential part of BSC schemes is the Move-to-Front (MTF) coder (recency ranking). In this talk we analyze the MTF coder and introduce a different coding (ranking) scheme, inversion coder. We prove the information theoretic relationship between interval ranks and canonical sorting permutations. Finally, we explore the relationship between inversion ranks and recency ranks and show that the inversion coding (ranking) technique introduced is superior to interval ranking and as well as recency ranking.

This is another in the Departmental Pizza Seminar series.

Wednesday, December 8th 2004, 5:00pm, Ungar 411

Craig Snoeyink
Computer Engineering
University of Miami

will present

Writing an Operating System From Scratch
(Not Necessarily a Good Idea)

The standard operating system class will give you a good introduction to how operating systems work. They will likely give you overviews of some of the basic components like scheduling, IPC, memory management, etc. The knowledge that you receive will give you some direction, but little practical assistance should you choose to write your own OS. This talk is intended to provide some insight as to the specific challenges of building an OS from scratch.

This is another in the Departmental Pizza Seminar series.

Friday, October 29th 2004, 4:30pm, Ungar 402

Michael Mascagni
Department of Computer Science
Florida State University

will present

On the Scrambled Halton Sequence

The Halton sequence is one of the standard (along with (t,s)-sequences and lattice points) low-discrepancy sequences, and thus is widely used in quasi-Monte Carlo applications. One of its important advantages is that the Halton sequence is easy to implement due to its definition via the radical inverse function. However, the original Halton sequence suffers from correlations between radical inverse functions with different bases used for different dimensions. These correlations result in poorly distributed two-dimensional projections. A standard solution to this is to use a randomized (scrambled) version of the Halton sequence. Here, we analyze the correlations in the standard Halton sequence, and based on this analysis propose a new and simpler modified scrambling algorithm. We also provide a number theoretic criterion to choose the optimal scrambling from among a large family of random scramblings. Based on this criterion, we have found the optimal scrambling for up to 60 dimensions for the Halton sequence. This derandomized Halton sequence is then numerically tested and shown empirically to be far superior to the original sequence.

This work is joint with Ms. Hongmei, and is part of her Ph.D. dissertation.

Monday, October 18th 2004, 4:00pm, Ungar 402

Sergei Nirenburg
University of Maryland

will present

Ontological Semantics: An Overview

The term ontological semantics refers to the apparatus of describing and manipulating meaning in natural language texts. Basic ontological-semantic analyzers take natural language texts as inputs and generate machine-tractable text meaning representations (TMRs) that form the basis of various reasoning processes. Ontological-semantic text generators take TMRs as inputs and produce natural language texts. Ontological-semantic systems centrally rely on extensive static knowledge resources:

• a language-independent ontology, the model of the world that includes models of intelligent agents;
• ontology-oriented lexicons (and onomasticons, or lexicons of proper names) for each natural language in the system; and
• a fact repository consisting of instances of ontological concepts as well as remembered text meaning representations.

Applications of ontological semantics include knowledge-based machine translation, information retrieval and extraction, text summarization, ontological support for reasoning systems, including networks of human and software agents, general knowledge management and others. In this talk I will give an overview of ontological-semantic processing and static resources and discuss its use in applications.

Wednesday, October 13th 2004, 5:00pm, Ungar 411

Geoff Sutcliffe
Department of Computer Science
University of Miami

will present

Standards and Tools for Component-based Automated Reasoning

The development and deployment of component-based automated reasoning systems relies on an adequate infrastructure of standards and tools that allow the component systems to execute and interact in a controlled and reliable way. In the field of 1st order automated reasoning there has been limited standardization across the available component systems, resulting in adhoc combination techniques and a limited range of general purpose tools. This work presents two emerging standards in the 1st order automated reasoning community, whose adoption is leading to greater compatibility between component systems. Six general purpose tools that conform to these standards are then described. The standards and tools facilitate direct communication between components of complex systems, providing seamless integration and greater reasoning productivity.

This is another in the Departmental Pizza Seminar series.

Wednesday, April 28th 2004, 3:00pm, Ungar 402

Vijay Srinivasan
IBM Corporation & Columbia University

will present

Theory of Dimensioning

Dimensions are numerical values assigned to certain geometric parameters. While dimensioning has been a common and basic engineering activity practiced over several centuries, a theory of dimensioning emerged only in the last decade of the 20th century. This theory is a synthesis of many ideas that range from elementary Euclidean geometry to some advanced Lie group classification of rigid motions. A prime motivator for this effort was the need to standardize the specification and exchange of product information in the computerized management of product lifecycle.

In this talk I will explain some elements of the theory of dimensioning and will point out how this also serves as a theory of parameterizing geometric models.

(This talk is based on the speaker's book "Theory of Dimensioning" published by Marcel-Dekker in October, 2003.)

About the Speaker: Vijay Srinivasan is the Program Manager for PLM Research, Standards, and Academic Programs at the IBM Corporation in New York. He is also an Adjunct Professor in the Engineering School at Columbia University, NY. He is a member of several national and international standards committees in the areas of product specification, verification, and data exchange.

Wednesday, April 21st 2004, 5:00pm, Ungar 411

Tony Pich
oasyssoft

will present

e-Business Designer: Web development made easy

Increasing competition in the development tool market has pushed vendors to combine their offerings into single products or sell them as suites. In the Web tool market, for instance, with the exception of some XML technologies, vendors are repackaging their tools into suites. eBD Software is offering alternative development methodologies to traditional Web tools such as Dreamweaver, introducing its e-Business Designer (eBD) tool in the United States.

eBD is a mix of portal and Web development environment that can generate portals, intranets and document management systems. It takes a radically different approach to Web Development. The tool follows a unique object model in which most of the features in the environment used for creating applications work like objects. Folders, pages and tables, for instance, all fall under a hierarchical structure and are treated as complete entities. Each of these structures contain a number of wizards so users can build fairly sophisticated portals with little knowledge of Web development. The wizards also come with features that enhance a site. With a few clicks, users can add calendaring, chatting and mailing links.

The talk will be a introduction to this new software solution, which will include a demo, real case studies and a discussion on this and similar technologies.

This is another in the Departmental Pizza Seminar series.

Wednesday, April 14th 2004, 5:00pm, Ungar 411

Geoff Sutcliffe
University of Miami

will present

How to Give a Successful Seminar

Almost everyone has to give a public presentation at some time. Graduate students have to defend their theses, research students have to present their results, and many jobs require presentations. It is common to have little or no experience when you give your first presentation, and you may even be a little nervous! This seminar is aimed at (graduate, research, and other) students. It describes how to present a successful seminar, in a simple standard format. This seminar does not try to teach general speaking skills, nor impose any personally preferred techniques. It covers the structure of a seminar, the use of visual aids, speaking technique, and how to cope with questions.

This is another in the Departmental Pizza Seminar series.

Wednesday, April 7th 2004, 5:00pm, Ungar 411

Steven Trac
Department of Computer Science
University of Miami

will present

Robust Arrangement Algorithms for
Lines and Semi-algebraic Curves

We present robust floating point algorithms for constructing arrangements of lines and semi-algebraic curves and for manipulating semi-algebraic sets. These robust algorithms construct a consistent combinatorial arrangement, and thus can handle every type of degeneracy (special cases that can give the algorithm problems). Due to floating point inaccuracies in computer arithmetic, degenerate cases arise. For example, three curves that meet at a point or two points with equal x coordinates can cause problems. An arrangement is a set of lines and curves that induces a subdivision of the plane, and it consists of vertices, edges, and cells. The construction of an arrangement is done using a plane sweep algorithm to find all the intersections. Our input is a set of continuous x-monotonic curves and lines delimited by endpoints. The sweep output is a partial vertical order on our set of segments, and these segments are obtained by splitting the input segments by their intersections. The sweep output is then converted to an arrangement (the segments are edges and their endpoints are vertices). Segments are grouped into loops by traversing through the vertices and edges, and these loops are grouped into cells using the segment order. We can now use our arrangement algorithm to rotate semi-algebraic sets and to perform set operations. The sweep runs in O((n + N + k) log n) where n is the number of curves/lines, N is the number of intersections, and k = O(n3) is the number of inconsistencies.

This is research done under the guidance of my advisor Dr. Victor Milenkovic.

This is another in the Departmental Pizza Seminar series.

Wednesday, March 31st 2004, 5:00pm, Ungar 411

Wei Zhou
Department of Computer Science
University of Miami

will present

Bandwidth Estimation for Multiplexed Videos Using
MMG-based Single Video Traffic Model

A Video on Demand (VoD) system is expected to transmit many movies over a single channel as demanded by the end users. When several videos are transmitted simultaneously over a link the effective bandwidth required per video is usually much lower than that needed by a single video because of multiplexing gain. Fast and accurate estimation of multiplexing gain is necessary for developing call admission control (CAC) algorithms. Known models which estimate queue size and effective bandwidth of multiplexed video system cannot capture frame size variations in different segments of a video, and are not very useful particularly when the number of videos is not too large.

The multinomial model proposed in this paper is built on a Markov-modulated gamma (MMG)-based traffic model of a single video. It takes into consideration average frame size variations in different segments of a video and can predict multiplexing gain for any number of multiplexed videos. The model has been validated using MPEG traces of commercial movies.

This is another in the Departmental Pizza Seminar series.

Wednesday, March 24th 2004, 3:00pm, Ungar 426

Dr. Luis E. Ortiz
Department of Computer and Information Science
University of Pennsylvania

will present

Algorithms for Graphical Games

A graphical game merges network models with traditional game theory to achieve a new compact representation better suited for large-population game theory. The model graph allows the explicit encoding of the detailed structure of strategic interactions yet has a natural and simple intuitive meaning: players are represented as nodes in the graph and the payoff of a player is only a function of the actions of the player's neighbors in the graph and those of the player itself.

In this talk, I will discuss computational and algorithmic aspects of graphical games. Algorithms for basic computations, including Nash and correlated equilibria, will be presented. Connections to related topics, such as graphical models for probabilistic modeling and inference, will be discussed. Throughout the talk, I will briefly sketch the state of the art in graphical games as well as recent work on extensions to related economic models. Time permitting, I will conclude with a discussion on open problems and future research directions.

This is joint work with Sham Kakade (Penn), Michael Kearns (Penn), John Langford (TTI-Chicago), Michael Littman (Rutgers), Nick Montfort (Penn), Robin Pemantle (Penn) and Siddharth Suri (Penn).

Wednesday, March 12th 2004, 5:00pm, Ungar 411

Burt Rosenberg
Department of Computer Science
University of Miami

will present

A Proposed Method for
Electronic Submission of Course Evaluations

University Course Evaluations are presently done on paper. To move this to electronic format, there are several issues similar to those of electronic voting, in particular, anonymity for the author, authenticity of the review, correct collection, and accountability.

The speaker takes a stab at designing such a system, and hopes to have his students implement a model.

Come prepared to break my system!

This is another in the Departmental Pizza Seminar series.

Friday, March 12th 2004, 5:00pm, Ungar 426

Stephen Kobourov
Department of Computer Science
University of Arizona

will present

Dynamic Data Visualization

Problems in information visualization often involve the display of a set of objects and their relationships, that can be modeled as graphs. While static graphs arise in many applications, dynamic processes give rise to graphs that evolve through time. Such dynamic processes can be found in software engineering, internet/telecommunications traffic, and social networks, among others. Typically, the underlying graph structures are large, and depending on the time-granularity, may contain from a few to a large number of timeslices. The graph representing a particular timeslice may contain fewer/more vertices/edges than the one representing the previous timeslice, or it may differ in some other graph attributes, such as node-weights, edge-weights, or labels. We describe algorithms and techniques for visualization of dynamic graph processes, using the evolution of the scientific literature as an example of such a process. In particular, we use a local copy of the ACM Digital Library of Scientific Literature dataset with over 100,000 authors and 100,000 papers.

Wednesday, March 3rd 2004, 5:00pm, Ungar 411

Victor Milenkovic
Department of Computer Science
University of Miami

will present

Optimization-Based Animation

Current techniques for rigid body simulation run slowly on scenes with many bodies in close proximity. Each time two bodies collide or make or break a static contact, the simulator must interrupt the numerical integration of velocities and accelerations. Even for simple scenes, the number of discontinuities per frame time can rise to the millions. An efficient optimization-based animation (OBA) algorithm is presented which can simulate scenes with many convex three-dimensional bodies settling into stacks and other "crowded" arrangements. This algorithm simulates Newtonian (second order) physics and Coulomb friction, and it uses quadratic programming (QP) to calculate new positions, momenta and accelerations strictly at frame times. Contact points are synchronized at the end of each frame. The extremely small integration steps inherent to traditional simulation techniques are avoided. Non-convex bodies are simulated as unions of convex bodies. Links and joints are simulated successfully with bi-directional constraints. A hybrid of OBA and retroactive detection (RD) has been implemented as well. A review of existing work finds no other packages that can simulate similarly complex scenes in a practical amount of time.

Harald Schmidl and I presented this work at SIGGRAPH 2001. He and I developed newer work as part of his Ph.D. at UM. I have continued to work on my own, and I am now working with Kanishka Bhaduri. This talk will start with an earlier 1996 SIGGRAPH paper, continue with the SIGGRAPH 2001 paper, and continue towards the present day as time allows.

This is another in the Departmental Pizza Seminar series.

February 25th 2004, 5:00pm, Ungar 411

Peter Mitchell

will present

Human Information and Systems Design

Failure to understand the limitations of the human being, as a link in any computer/automated system in which the person interacts, will most certainly lead to significant problems in the operation of the system that will either: (a) result in safety problems that could harm or kill, (b) cost the organization large amounts of money to correct problems, (c) lead to unhappy employees who are forced to use a system that is awkward and cumbersome. All of these outcomes are unacceptable. There exists today, a scientific, proven technique, that enable us to analyze the needs of the user/system, to study the environment within which the system is being developed, and formulate a user interface design specification which will ensure that these problems do not occur.

Peter Mitchell is an expert in Human-computer Interaction and Usability Testing. He is President Emeritus of the New York City Chapter of the Usability Professionals' Association, and a part time lecturer in the Department of Industrial Engineering. He owns the company ergo9.

This is another in the Departmental Pizza Seminar series.

Department of Computer Science Colloquium

Geoff Sutcliffe, Diego Belfiore
University of Miami

will present

A Derivation Verifier of the Semantic Kind

Automated Theorem Proving (ATP) systems are complex pieces of software, and thus may have bugs that make them unsound or incomplete. The derivations output by an ATP system may be semantically verified by trusted systems that check the required semantic properties of each inference step. Such verification may need to be augmented by structural verification that checks that inferences have been used correctly in the context of the overall derivation. This talk describes the design and implementation of the derivation verifier DVDV, which uses semantic verification and does structural checking for various situations.

Wed Feb 18, 5:00
In Ungar, Room 402
Free Pizza

Received: Sun Feb 1 0:17:18 2004

Department of Computer Science

Colloquium

Yury Puzis, Geoff Sutcliffe, Yuan Zhang and Yi Gao
University of Miami

will present

Proving hard theorems in rich theories, part II

A Rich Theory is one whose axioms (expressed in 1st order logic) contain a large number of predicates and functors, and whose theorems are often provable from a subset of the axioms. Experimental results have shown that the removal of just a few unnecessary axioms can have a significant effect on the performance of an ATP system on a problem. This work introduces a relevance measure that estimates the potential usefulness of an axiom for proving a given theorem. The measure has been used in a control algorithm that selects combinations of axioms based on their relevance, to form {\em axiom reduced} problems that are submitted to an ATP system. Evaluation shows that this system performs better than the ATP system alone. Additionally, a scheme for using the relevance measures for selecting lemmas to augment the axioms of a problem is introduced.

Monday, November 17, 5:00 PM
Either Ungar 402 or 506, as scheduling permits.
Pizza and drinks are provided by the department.

Department of Computer Science

Colloquium

Victor J. Milenkovic
University of Miami

will present

Robust Set Operations on Semi-Algebraic Sets in the Plane

At my last pizza talk, I described how to robustly construct an arrangement of algebraic curves. The ultimate purpose was CAD: computer aided design. However, constructing arrangements is still a far cry from designing stronger bridges, faster chips, or bigger SUVs. This talk will show how to define sets in the plane bounded by segments of algebraic curves and how to apply rigid transformations and how to perform set operations on them: union, intersection, symmetric difference, etc. Most of the work goes into dealing with degenerate cases: unexpected coincidences''. An important trick is a spin on the wheel of fortune'': apply a random rotation to the set before attempting the operation.

Again, this is joint work with Elish Sacks of Purdue University.

Monday, November 10, 2003, 5:00 PM
Either Ungar 402 or 506, as scheduling permits.
Pizza and drinks are provided by the department.

Department of Computer Science

Colloquium

Brian Coomes
University of Miami

will continue his presentation

Floating Point Computations

This is the third talk by Brian on this subject. Analysis of floating point error of the dot product was given in the previous talk, and error for various matrix norms.

Monday, November 3, 2003, 5:00 PM
Either Ungar 402 or 506, as scheduling permits.
Pizza and drinks are provided by the department.

Received: Fri Oct 31 10:46:59 2003

Department of Computer Science

Colloquium

Brian Coomes
University of Miami

will present

IEEE754 Floating Point Arithmetic, continue

We will show how to use the floating-point arithmetic specified in IEEE754 to produce rigorous mathematical estimates of various quantities.

See seminar, 13 October for further details.

Monday, 27 October 5:00 PM
In Ungar 402 or 506
Pizza is served at 5:00 PM
All are welcome.

Received: Fri Oct 17 18:05:50 2003

Department of Computer Science

Colloquium

Burton Rosenberg
University of Miami

will present

Secure computation with a deck of cards

A short presentation in which an ordinary deck of cards is used to compute an arbitrary boolean function by two mistrustful parties. By encoding bits and cutting the deck, neither party will learn the other party's input except for what is implied by the function output itself.

Monday, Oct 20, 5:00 PM
In Ungar, Room CC402 or 506, depending on weather.
(Weather what? whether or not it is in 402, of course).
Pizza will be served at 5:00 PM.

Received: Fri Oct 17 18:00:58 2003

Department of Computer Science

Colloquium

Brian Coomes
University of Miami

will present

IEEE754 Floating Point Arithmetic
and Rigorous Error Estimates

Before the approval of the IEEE754 standard in 1985, many computers had idiosyncratic or problematic floating-point implementations. This led to the discovery of counterintuitive, even bizarre, programmatic situations, which in turn created a general aura of distrust for floating-point arithmetic. The standard, when implemented correctly, gives cross-platform consistency and predictability to floating-point arithmetic. While the aura of distrust lingers, the general status of floating-point arithmetic has improved. We will show how to use the floating-point arithmetic specified in IEEE754 to produce rigorous mathematical estimates of various quantities. As an example, we will show how to use floating-point arithmetic to compute a rigorous upper bound for the Euclidean norm of the inverse of a square matrix approximated by a matrix of floating-point numbers.

Monday, 13 October, 5:00 PM
In Ungar, Room CC402
This the 30th in our series of Pizza seminars.
Pizza is provided.

Received: Sun Oct 12 22:03:53 2003

Department of Computer Science

Colloquium

Chip Elliott
BBN, Boston

will present

Quantum Cryptography in Practice

Abstract:
BBN, Harvard, and Boston University are building the DARPA Quantum Network, the world's first network that delivers end-to-end network security via high-speed Quantum Cryptography, and testing that Network against sophisticated eavesdropping attacks. The first network link has been up and steadily operational in our laboratory since December 2002, and will soon be built out across the metro Boston area via standard telecom (dark) fiber. In this talk, we introduce quantum cryptography, discuss its relation to modern secure networks, and describe its unusual physical layer, its specialized quantum cryptographic protocol suite (quite interesting in its own right), and our extensions to IPsec to integrate it with quantum cryptography.

Monday, October 6, 5:00 PM
Ungar Room 506
This is the 29th Pizza Seminar. Pizza is served.
Received: Mon Sep 22 19:12:56 2003

Department of Computer Science

Colloquium

Burton Rosenberg
University of Miami

will present

Towards Implementing Secure 2-Party Computation

I describe work with students Yi Gao, Yuan Zhang and Vera Arsova to implement the distributed cryptographic evaluation of functions.

We are concerned with the two party case. X and Y have values x and y and wish to collectively compute some function f(x,y). At the end of the computation X learns nothing of the value y except what can be inferred from x and the computed f(x,y). Likewise for Y. The protocol must also not release additional information even if Y cheats on the protocol.

Background the the problem will be followed by a discussion of the implementation and a demonstration.

Monday, 29 September, 5:00 PM
In Ungar, Room 402
This is the 28th Pizza Seminar. Pizza and drinks are served.
Received: Mon Sep 22 19:02:43 2003

Monday, October 6, 2003, 5:00pm, Ungar 402

Chip Elliott
BBN, Boston MA.

Quantum Cryptography in Practice

BBN, Harvard, and Boston University are building the DARPA Quantum Network, the world's first network that delivers end-to-end network security via high-speed Quantum Cryptography, and testing that Network against sophisticated eavesdropping attacks. The first network link has been up and steadily operational in our laboratory since December 2002, and will soon be built out across the metro Boston area via standard telecom (dark) fiber. In this talk, we introduce quantum cryptography, discuss its relation to modern secure networks, and describe its unusual physical layer, its specialized quantum cryptographic protocol suite (quite interesting in its own right), and our extensions to IPsec to integrate it with quantum cryptography.

This is the 27rd in the Departmental Pizza Seminar series.

Monday, September 22nd 2003, 5:00pm, Ungar 402

Dr. Christian Duncan
Department of Computer Science
University of Miami

Topics in Graph Drawing:
Simultaneous Visualization of Multiple Graphs

The drawing of a graph onto a medium such as a computer screen is a fundamental yet challenging problem in computer science. The algorithms developed incorporate techniques from computational geometry, graph theory, optimization, and computer graphics.

In general, the underlying problem is to take a graph G composed of vertices (V) and edges (E) between the vertices and place (embed) these vertices and edges on the screen in a visually appealing and informative manner. The precise definition of appealing'' and informative'' are of great debate in the graph drawing community but certain standards are almost universally accepted. For example, graphs with multiple edge crossings are considered less appealing and more confusing than similar graphs without any crossings.

In this talk, we shall introduce the area of graph drawing, give a brief description of basic results and problems in the field, and discuss some of our own recent results in the area of simultaneous graph embeddings.

This is the 25rd in the Departmental Pizza Seminar series.

Monday, September 15th 2003, 5:00pm, Ungar 402

Dr Victor Milenkovic
Department of Computer Science
University of Miami

An Inconsistency Sensitive Arrangement Algorithm for Curve Segments

We present a robust arrangement algorithm for algebraic curves based on floating point arithmetic. The algorithm performs a line sweep, tests the consistency of each sweep update, and modifies the input to prevent inconsistent updates. The output arrangement is realizable by semi-algebraic curves that are close to the input curves. We present a new performance model for robust computational geometry in which running times and error bounds are expressed in terms of the number of input inconsistencies. An inconsistency is a combinatorial property that is derivable with a given set of numerical algorithms, but that is not realizable. The running time of the arrangement algorithm is O((n+N)\log n+k(n+N)\log n) for n curves with N intersection points and with k=O(n3) inconsistencies. The distance between the realization curves and the input is O(ε+knε) where ε is the curve intersection accuracy. The output size is always the standard O(n+N). We show experimentally that k is zero for generic inputs and is tiny even for highly degenerate inputs. Hence, the algorithm running time on real-world inputs equals that of a standard sweep and the realization error equals the curve intersection error.

Joint work with Elisha Sacks of Purdue University.

This is the 24rd in the Departmental Pizza Seminar series.

Monday, September 8th 2003, 5:00pm, Ungar 402
Tuesday, September 9th 2003, 5:00pm, Ungar 402
Wednesday, September 10th 2003, 5:00pm, Ungar 406

Joe Marcus Clarke
Cisco Systems

The Marcus Trilogy

Joe Clarke gives two talks on FreeBSD and workshop on firewalls. The first talk will be given as part of the Computer Science Pizza Seminar series. The next two presentations will be given on the following two days.

The FreeBSD Way.
Monday, September 8th, 2003

This talk will discuss the origins and direction of FreeBSD. We will cover the political and social layouts of the project. We will discuss the philosophy of the operating system and how it is developed, as well as the underlying BSD license. We will compare to Linux and GPL where appropriate. We will also discuss FreeBSD strengths and weaknesses, and show where FreeBSD fits into both the server and desktop niches.

Getting to know and love FreeBSD
Tuesday, September 9th, 2003

This talk will cover getting started and learning to use FreeBSD. Topics will include a basic overview of installing the operating system, and where to find help to get you to "the next step." We will discuss the layout of the file system as well as how system configuration is done in the FreeBSD world. We will go into great detail about the FreeBSD ports collection, and show how to find ported software as well as install and manage ports and packages. Finally, we will look at some advanced topics like tracking -STABLE and -CURRENT branches, and how to keep your sources up-to-date.

Working with the Cisco Pix firewall
Wednesday, September 9th, 2003

This workshop is designed to familiarize students with the workings of the Cisco PIX firewall. We discuss terminology as well as configurations. We will be configuring a sample network using a PIX 501, and then demonstrating how to troubleshoot problems in the network.

Joe Marcus Clarke, Bio.

Joe Marcus Clarke holds a Bachelor of Science in Computer Science from the University of Miami. Since graduating, he has worked for Cisco Systems, Inc. in the Technical Assistance Center (TAC) supporting Cisco's network management software and technologies. He is Cisco Certified Internetwork Expert #5384 and is also a Sun Certified Java Programmer. He is a ports committer for the FreeBSD project specializing in GNOME on FreeBSD. He also sits on the Port Managers team for FreeBSD. Other Open Source projects he works on include Netatalk and the Cisco Open Source Initiative (COSI).

This is the 23rd in the Departmental Pizza Seminar series.

April 21st 2003, 5:00pm, Ungar 411

Dr Geoff Sutcliffe
and
Ms Yi Gao
Department of Computer Science
University of Miami

Proving Hard Theorems in Rich Theories

A Rich Theory is one whose axioms (expressed in 1st order logic) contain a large number of predicates and functors, and whose theorems are often provable from a subset of the axioms. Examples of rich theories are set theory, geometry, and homological algebra. The complexity and scope of rich theories contribute to the difficulty of finding proofs of theorems using Automated Theorem Proving (ATP) systems. This seminar describes techniques that have been developed to improve the performance of ATP systems on theorems in rich theories.

This is the 22nd in the Departmental Pizza Seminar series.

April 14th 2003, 5:00pm, Ungar 411

Dr Debasish Dev Roy
Department of Computer Science
University of Miami

Protocol, Architecture, and Planning Issues for
Video and Multimedia Services in
Broadband Wireline Communication Networks

In this talk, the presenter will discuss today's and tommorrow's emerging consumer applications/services based on video and multimedia communications. The scope of this tutorial-style presentation will include application/service classes, common technology, protocol and architectures and network planning issues for supporting video and multimedia applications over Broadband wireline communication networks. Specific discussion will be based on practical issues related to key network planning parameters such as application domains, bandwidth, quality-of-service (QoS) and multicast.

This is the 21st in the Departmental Pizza Seminar series.

April 7th 2003, 5:00pm, Ungar 411

Dr Burton Rosenberg
Department of Computer Science
University of Miami

The AKS Proof of Primes in P

This summer Agrawal, Kayal and Saxena solved the long outstanding problem of a P-time algorithm for deciding whether a given integer is prime. In this pizza seminar I will present their proof.

This is the 20th in the Departmental Pizza Seminar series.

March 31st 2003, 5:00pm, Ungar 402

Dr Colby Leider
Music Engineering
School of Music
University of Miami

The Human Gesture as Music:
Real-Time Control of Sound Synthesis and Processing Parameters

Throughout the twentieth century, composers, performers, and technologists have yearned for robust control interfaces to mediate between musician and machine. Recent developments have enabled the creation of inexpensive gestural interfaces that facilitate straightforward, meaningful, and precise mapping of gesture into sound. I will introduce the notion and importance of the "alternate controller" and discuss several of my own instruments and interfaces. Plenty of sound will be served.

This is the 19th in the Departmental Pizza Seminar series.

March 24th 2003, 5:00pm, Ungar 411

Dr Debasish Dev Roy
Department of Computer Science
University of Miami

Applications and Services for Commercial Deployment of
Third Generation (3G) Mobile Wireless Communication (Cellular) Systems

In this talk, the presenter will discuss today's and tomorrow's emerging consumer applications and services which will drive widespread deployment and acceptance of the Third Generation (3G) Mobile Wireless Communication (Cellular) Systems, blurring out the long debated gray line between myth and reality of 3G. Specific discussion will be based on key parameters of popular services/application, resulting impact/challenges on 3G system implementation, technology enablers and service migration from 2G to 3G Networks.

This is the 18th in the Departmental Pizza Seminar series.

March 17th 2003, 5:00pm, Ungar 411

Dr Victor Milenkovic
Department of Computer Science
University of Miami

Densest Translational Lattice Packing of Non-convex Polygons

A translation lattice packing of k polygons P1, P2, P3,...,Pk is a (non-overlapping) packing of the k polygons which is replicated without overlap at each point of a lattice i0v0+i1v1, where v0 and v1 are vectors generating the lattice and i0 and i1 range over all integers. A densest translational lattice packing is one which minimizes the area |v0 x v1| of the fundamental parallelogram. An algorithm and implementation is given for densest translation lattice packing. This algorithm has useful applications in industry, particularly clothing manufacture.

This is the 17th in the Departmental Pizza Seminar series.

March 3rd 2003, 5:00pm, Ungar 411

Dr Victor Milenkovic
Department of Computer Science
University of Miami

Cleats on Ice

A new physical model, Cleats on Ice (CoI), is described for calculating the acceleration of many rigid bodies with simultaneous contacts under the forces of gravity and contact friction. Under the CoI model, iterated quadratic programming (QP) suffices to calculate the accelerations. Like computationally more intensive complementarity-based techniques, CoI ensures complementarity: at each contact, a pair of bodies press on each other or accelerate away from each other but not both. However, CoI does so in an implicit rather than explicit manner that is both scalable and easy to implement. Furthermore, CoI can take advantage of powerful, well-maintained, and well-documented commercial optimization libraries. As part of a larger simulation package, CoI can practically generate realistic animations of scenes with over 200 bodies and over 500 simultaneous contacts.

Joint work with Harald Schmidl.

This is the 16th in the Departmental Pizza Seminar series.

February 24th 2003, 5:00pm, Ungar 411

Dr Geoff Sutcliffe
Department of Computer Science
University of Miami

Thousands of Solutions from Theorem Provers

The TPTP (Thousands of Problems for Theorem Provers) Problem Library is a library of test problems for Automated Theorem Proving (ATP) systems for 1st order classical logic. Since the first release of the TPTP in 1993, many researchers have used the TPTP as an appropriate and convenient basis for testing their ATP systems. The TPTP is now the de facto standard for testing and evaluating ATP systems. The TSTP Solution Library is the flip-side of the TPTP. The TSTP will maintain solutions to all the problems in the TPTP, from all currently available state-of-the-art ATP systems. The solutions output by the systems will be translated and stored in a common format, so that the TSTP can be consistently interrogated and analysed. It is expected that the TSTP will immediately contribute to the development of ATP systems, by allowing system developers to easily access and analyse other systems' solutions to TPTP problems. In a broader context the TSTP will provide a basis for understanding and comparing proof structures, will be used for certification of ATP systems, and form a framework within which users will be able to create and solve more problems using ATP tools and techniques.

This talk describes the work in progress towards the TSTP, including motivations, requirements, design decisions, and the various projects that are contributing to and using the TSTP.

This is the 15th in the Departmental Pizza Seminar series.

January 21st 2003, 4:00pm, Ungar 506

Dr Daniel G. Aliaga
Princeton University

Capturing and Rendering Real-World Environments

Computer simulation of real-world environments is one of the grand challenges of computer graphics. Applications for this technology include remote education, virtual heritage, specialist training, electronic commerce, and entertainment. Unfortunately, current computer graphics techniques fall far short of providing solutions for this challenge. In this presentation, I present a new approach to capturing and rendering large real-world environments, including several successful demonstrations and future research plans.

My ultimate goal is to allow an untrained operator to walk into a city or building (e.g., a museum) and wave around some device that captures a digital model, which later can be used to provide many people the realistic visual experience of "walking" through the environment interactively. My approach is to take advantage of recent technology trends and to obtain a dense and automatic sampling of a large viewpoint space with omnidirectional images. This strategy replaces the difficult computer vision problems of 3D reconstruction and surface reflectance modeling with the easier problems of motorized cart navigation, data compression, and working set management.

I also provide a research plan to capture and reconstruct large environments as well as a summary of previous approaches. This plan benefits from collaborative efforts in robotics (for building self-navigating high-resolution capture devices), computer vision (for developing image-reconstruction algorithms), and systems (for building and deploying large software systems over a network) and from developing applications to foment interactive tourism, to preserve historical sites, and to assist with simulation and training scenarios.

January 24th 2003, 4:00pm, Ungar 402

Dr James Abello
DIMACS
Rutgers University

Massive Graph Mining

A variety of massive data sets exhibit an underlying structure that can be modeled as dynamic weighted multi-digraphs. Their sizes range from tens of gigabytes to petabytes. These include the World Wide Web, Internet Traffic and Telephone Call Detail. These data sets sheer volume brings with it a series of computational and visualization challenges due mainly to the I/O and Screen Bottlenecks.

We present external memory algorithms for connectivity and minimum spanning trees together with heuristics for quasi-clique finding. We describe how hierarchy trees help us to cope in a unified manner with both, the I/O and screen bottlenecks. This line of research has suggested the need to look for "novel" graph representations in order to provide a user or a data mining engine with informed navigation paths.

We will discuss our results with graphs having on the order of 200 million vertices and several billion edges and we will point out some mathematical problems that have surfaced along the way.

The overall goal is to extract useful information that can be brought into a user's palm top and to export these techniques to other mining domains.

Information about some of James's current visualization research projects can be obtained by accessing www.research.att.com/areas/visualization and http://www.visdays.com/.

November 18th 2002, 4:00pm, Ungar 402

Dr Barak A. Pearlmutter
Department of Computer Science & Department of Neurosciences
University of New Mexico

Blind Separation and Fast Localization for MEG

The strength of magnetoencephalography (MEG) is that magnetic fields are not smeared by the skull, leading to greater potential spatial sensitivity. Unfortunately, this penetrative ability of magnetic fields also causes each sensor to receive signals from a large number of active sources and makes it difficult to shield out noise, both external (power grid) and internal (heartbeat, eyeblinks). We have made a systematic effort to improve MEG's performance through new signal processing for separation and dipole localization, both raising the effective performance of current MEG systems and relaxing constraints that have constrained MEG hardware design.

For separation, we applied SOBI to complex MEG data. This segregated non-neuronal sources from neuronal ones, and neuronal ones from each other. The separated neuronal sources seem focal, and have temporal responses consistent with their estimated locations. We can routinely isolate sources within modalities, across modalities, across tasks, and across subjects. The SNR is high enough to allow estimation of response onset times for single trials, which can be measured in visual, auditory, and somatosensory modalities with detection rates over 95%. Combined with an improved ability to localize the underlying neuronal sources, this makes possible the non-invasive study of a range of perceptual and memory functions that depend upon the timing of neuronal events occurring in specific brain regions.

With many recovered sources, dipole localization becomes a bottleneck. For both a 4D Neuroimaging Neuromag-122 and the experimental LANL SiS Mark I with superconducting magnetic mirrors, we addressed the single dipole localization problem at low SNR, achieving a reduction in localization time from 449ms to 0.5ms at an accuracy of 12mm, and to 35ms at an accuracy of 2.8mm. Our fast fully automatic noise-robust localizer is suitable for real-time applications, such as closed-loop experimental protocols and brain-computer interfaces.

Publications: http://www.cs.unm.edu/~bap/publications.html

November 11th 2002, 5:00pm, Ungar 411

Dr Debasish Dev Roy

End-to-End QoS in IP-based Multimedia Networks

In this presentation, I will cover the concepts of QoS, challenges in IP-based Multimedia networks, overall architecture, present state of design, implementation and related issues.

This is the 14th in the Departmental Pizza Seminar series.

November 8th 2002, 10:00am, Ungar 411

Prof. L. M. Patnaik
IISc., Bangalore, India

Mobile Computing: Trends and Challenges

Recent years have witnessed significant growth and interest in wireless communication. Coupled with the advances in compter hardware and architecture, progress in wireless communication has led to significant interests in the area of Mobile Computing, a paradigm of computing anytime, anywhere. In this talk we present a synthesizing overview of several issues related to architecture, mobile host detection, need for low power and large bandwidth, etc. Challenges involved in designing such systems will also be highlighted.

November 4th 2002, 5:00pm, Ungar 411

Dr Huseyin Kocak
University of Miami

That Picture is Almost Certainly Bogus!
An Illustrated Introduction to Chaotic Numerics

For detailed analyses of specific chaotic systems one must often resort to numerical simulations. Unfortunately, chaotic systems amplify small errors at an exponential rate, which makes their simulations necessarily unreliable. Indeed, many published pictures of solutions of chaotic systems are unintentional forgeries. In this talk we will present several striking simulations of chaotic systems using the soon-to-be-released Phaser (www.phaser.com). We will then reveal what these bogus pictures may really depict.

This is the 13th in the Departmental Pizza Seminar series.

October 21st 2002, 5:00pm, Ungar 411

Dr Burt Rosenberg
University of Miami

Optimal Exercise of Russian Options in the Binomial Model

The Russian Option is a two-party contract which creates a liability for the option seller to pay the option buyer an amount equal to the maximum price attained by a security over a specific time period, discounted for the option's age. For an N+1 step time period 0, 1, 2, ..., N, the option seller's liability at time step n, 0 ≤ n ≤ N, is,

L(n) = βn max0 ≤ t ≤ n st

where st is the security price at time t and β is the discount factor. In this work we consider the value of this option under a standard binomial model of security prices, and give efficient algorithms for value calculation.

This work gives an O(N2) dynamic programming algorithm determining the option price at all time steps as well as the optimal execution strategy.

This is the 12th in the Departmental Pizza Seminar series.

October 7th 2002, 5:00pm, Ungar 411

Dr Giri Narasimhan
Florida International University

Introduction to Bioinformatics

After introducing some basic molecular biology, this talk will present a sampling of problems in Bioinformatics where computer scientists have been involved and have left their mark.

This is the 11th in the Departmental Pizza Seminar series.

May 28th 2002, 4:00pm, Ungar 426

Dr Ofer Melnik
DIMACS Center, Rutgers University

Interpreting Black-box Models

A computational model is black-box if it is used without a clear understanding of how it works. A model may be black-box because its computation is hidden, or it may be black-box because its computation is hard to interpret. Examples of this latter category are semi-parametric and non-parametric models such as neural networks, nearest neighbor classifiers, and ensemble classifiers like boosting. Having greater interpretability for these models is important for advancing our understanding of how they work, but also necessary for verifying their functionality for real-world applications. In this talk I will present a new approach to analyzing the generalization strategies of these kinds of models by examining their decision region structure.

In the first half we will examine the decision region properties of feed-forward neural networks. Using the Decision Intersection Boundary Algorithm, we will look at noise in networks, good and bad generalization and learning.

In the second half we will build on the intuition gained from neural networks and introduce Decision Region Connectivity Analysis, a method that allows the computationally efficient analysis of classifiers in a model and dimensionality independent manner. I will demonstrate how DRCA can be used to understand and compare completely different types of high-dimensional classifiers (a 64-input OCR example), amongst its applications, and conclude with current research directions.

April 24th 2002, 5:00pm, Ungar 411

Dr Uttam Sarkar
University of Miami

Modeling Long-duration Variable Bit Rate (VBR) Video using
Markov-modulated Gamma-based Framework

Video traffic is one of the major sources of broadband network traffic. Because of large bandwidth requirement of high-quality uncompressed video most video are compressed using some variants of MPEG encoding before transmission. These compression algorithms can provide very high compression ratio while maintaining good quality of decompressed video. However, MPEG-coding provides different amount of compression for different frames and results in Variable Bit Rate (VBR) data known as VBR video. Accurate traffic models of VBR video are necessary for prediction of performance of any broadband network during its design and operation. We propose a frame size model of VBR video. It generates frame sizes for full-length VBR videos while preserving both GOP-periodicity and video-segment transitions.

A two-pass algorithm for analysis of a long-duration VBR video is presented. The algorithm identifies and partitions classes of video segments. Frames in each segment class produce three data-sets one each for I-, B-, and P-type frames. Each of these data-sets is modeled with a Gamma distribution. Markov processes model video segment transitions. We have used QQ plots to show visual similarity of model-generated VBR video data-sets with original data-set. Leaky-bucket simulation study is used to show similarity of data and frame loss rates between model-generated VBR videos and original video. Our study of frame-based VBR video also reveals even a low data loss rate could affect a large fraction of I- frames causing a significant degradation of the quality of transmitted video.

This is the 10th in the Departmental Pizza Seminar series.

April 17th 2002, 5:00pm, Ungar 411

Morgan Johnson
University of Miami

Identify Epileptic Seizure in EEG using Approximate Entropy

It has been shown that methods such as Correlation Dimension provide inconclusive results when determining the complexity/dynamics of transient EEG signals under the presence of multiple signal components. We propose the use of Approximate Entropy since it has proven to be robust to both signal noise and statistical outliers.

This is the 9th in the Departmental Pizza Seminar series.

April 5th 2002, 3:30pm, Ungar 426

Harald Schmidl
Interdepartmental Studies, University of Miami

Dissertation Defense

Optimization-Based Animation

A new paradigm for rigid body simulation is presented and analyzed. Current techniques for rigid body simulation run slowly on scenes with many bodies in close proximity. Each time two bodies collide or make or break a static contact, the simulator must interrupt the numerical integration of velocities and accelerations. Even for simple scenes, the number of discontinuities per frame time can rise to the millions. An efficient optimization-based animation (OBA) algorithm is presented which can simulate scenes with many convex three-dimensional bodies settling into stacks and other "crowded" arrangements. This algorithm simulates Newtonian (second order) physics and Coulomb friction, and it uses quadratic programming (QP) to calculate new positions, momenta, and accelerations strictly at frame times. The extremely small integration steps inherent to traditional simulation techniques are avoided.

Contact points are synchronized at the end of each frame. Resolving contacts with friction is known to be a difficult problem. Analytic force calculation can have ambiguous or non-existing solutions. Purely impulsive techniques avoid these ambiguous cases, but still require an excessive and computationally expensive number of updates in the case of many simultaneous contacts. It is shown informally that even taking into account advances in stiff integration techniques, penalty force methods cannot overcome this issue of running time in highly crowded scenes. New algorithms are presented that calculate simultaneous impulses to resolve collisions and static contacts under the Coulomb friction model. The simultaneous impulses are the solution to a QP.

In addition, the algorithms apply "bouncing at distance" and "freezing of bodies" to further speed up the simulation. These new QP algorithms are hybridized with a traditional priority queue momentum update scheme to allow sequential impulses when they are required for realism, such as in the office toy pendulum. When added to the implementation of OBA, these new algorithms increase the speed of the simulation by a factor of up to 30.

The position update has been hybridized with retroactive detection (RD) to prevent fast and thin bodies from passing through each other. Due to the modular design of the OBA simulator, the described techniques can be used as components in any existing simulator that follows a modular design of position update, finding contacts, and resolving contacts. Non-convex bodies are simulated as unions of convex bodies. Links and joints are simulated with bi-directional constraints. Analysis of the algorithm and discussion of example simulations is provided.

March 27th 2002, 5:00pm, Ungar 411

Dr Burton Rosenberg
University of Miami

Kickstart to Hacking the Computer Science Security Architecture

We discuss the security architecture for the Computer Science department. We review the goals and challenges of the environment, look at the technology used to respond to the challenges, and get into some detail about the proposed solution.

This is the 8th in the Departmental Pizza Seminar series.

March 22nd 2002, 4:00pm, Ungar 506

Dr Juliana Freire
Bell Labs

From XML Schema to Relations: A Cost-Based Approach to XML Storage

As applications manipulate an increasing amount of XML, there is growing interest in storing XML data in relational databases. Due to the mismatch between the complexity of XML's tree structure and the simplicity of flat relational tables, there are many ways to store the same XML document in an RDBMS, and a number of heuristic techniques have been proposed. These techniques typically define fixed mappings and do not take application characteristics into account. However, a fixed mapping is unlikely to work well for all possible applications.

LegoDB is a cost-based XML storage mapping engine that explores a space of possible XML-to-relational mappings and selects the best mapping for a given application. LegoDB leverages current XML and relational technologies: 1) it models the target application with an XML Schema, XML data statistics, and an XQuery workload; 2) the space of configurations is generated through XML-Schema rewritings; and 3) the best among the derived configurations is selected using cost estimates obtained through a standard relational optimizer. In this talk, I will describe the LegoDB storage engine and present experimental results that demonstrate the effectiveness of this approach.

Joint work with Philip Bohannon, Prasan Roy and Jerome Simeon.

March 21st 2002, 9:00am, Ungar 426

Dr Claudio T. Silva
AT&T Labs-Research

Direct Volume Rendering Techniques for Unstructured Grids

The need to visualize unstructured volumetric data arises in a broad spectrum of applications including structural dynamics, structural mechanics, thermodynamics, fluid mechanics, and shock physics. In this talk, I will focus on direct volume rendering, a term used to define a particular set of rendering techniques which avoid generating intermediary (surface) representations of the volume data. Instead, the scalar field is generally modeled as a cloud-like material, and rendered by computing a set of lighting equations.

I will present recently developed techniques for direct volume rendering of unstructured grids. First, I will talk about "projective methods". Direct volume rendering based on projective methods works by projecting, in visibility order, the polyhedral cells of a mesh onto the image plane, and incrementally compositing the cell's color and opacity into the final image. By combining a fast visibility ordering algorithm with extensive use of graphics hardware, it is possible to achieve real-time rendering of moderate size datasets with such techniques.

Then, I will present the ZSWEEP algorithm, which can be seen as an "extended" projective technique under a generalized hardware model. I will describe several features and (parallel, distributed, and out-of-core) extensions of the ZSWEEP algorithm. Because ZSWEEP is not limited by the amount of available main memory, it is suitable for rendering large data.

This work was done in collaboration with Joao Comba (Stanford/UFRGS), R. Cook (Lawrence Livermore National Labs), R. Farias (Mississippi State), J. Klosowski (IBM Research), N. Max (LLNL), J. Mitchell (SUNY, Stony Brook), and P. Williams (LLNL).

March 20th 2002, 5:00pm, Ungar 411

Dr Victor J. Milenkovic
University of Miami

Compaction and Overlap Minimization

COMPACTION takes a set of "loosely packed" geometric objects and packs them "as tightly as possible." It is akin to sitting on a suitcase to get it to close. Compaction has three inputs: 1) a set of non-overlapping geometric objects, such as polygons or polyhedra; 2) allowable transformations of those objects, such as translations or rotations; 3) an objective function to be minimized, such as gravitational energy or the z-coordinate of the "suitcase lid". Compaction plans a simultaneous non-overlapping motion of the objects that moves them to a local minimum of the objective function.

OVERLAP MINIMIZATION takes overlapping objects as inputs and minimizes a measure of their overlap.

Algorithms are presented for compaction and overlap minimization of polygons, "circular" polygons (circular arcs allowed in the boundary), and polyhedra. These algorithms are based on linear programming or quadratic programming. They also use the MINKOWSKI SUM or the MAXIMAL CONVEX SUBSET.

Some of these algorithms are widely used in the clothing industry or are the basis for new techniques for computer animation. They also are an important part of other algorithms for PACKING or other GLOBAL minimization problems.

This is the 7th in the Departmental Pizza Seminar series.

March 6th 2002, 5:00pm, Ungar 411

Sivaramakrishna Mopati
University of Miami

Call Admission Control Algorithm for Wireless Networks Using Fuzzy Logic

Historically, wireless networks have been designed primarily to extend the domain of circuited switched telephone service over cellular telephony. However, wireless networks that support data have now begun to appear. The third and fourth generation wireless networking technologies and wireless systems are being designed to support packet data and multimedia services in addition to voice. A typical wireless network has to support multiple classes of users with different traffic characteristics and QoS (Quality of Service) requirements. The measure of QoS is the probability of forced termination (Pft) of a call that was allowed to access the network. Guaranteeing the promised QoS for all admitted users while maximizing network efficiency is one of the major challenges. Call Admission Control plays a key role in achieving promised QoS for the mobile terminals (MTs) in a given network. Two previous CAC algorithms i) Guard Channel Scheme ii) Call Admission Control with call Pre-blocking are studied which aim to decrease Pft. In this work, we study both these schemes through extensive simulation and it is found that although they provide required QoS for constant traffic, they fail to adjust to the dynamic nature of the system and sometimes underutilize the network resources. We then propose a new Call Admission Control Algorithm that uses Fuzzy logic to adjust to the changing traffic parameters and provide any level of QoS to the users. It is also shown that computational burden on such system is very minimum.

This is the 6th in the Departmental Pizza Seminar series.

March 1st 2002, 4:00pm, Ungar 506

Dr. Inna Pivkina
University of Kentucky

Revision Programming

Revision programming is a knowledge representation formalism for describing constraints on databases, knowledge bases, and belief sets, and providing a computational mechanism to enforce them. Constraints are represented by sets of revision rules. Revision rules could be quite complex and are usually in a form of conditions (for instance, if these elements are present and those elements are absent, then this element must be absent). In addition to modeling logical constraints, revision rules specify a preferred way to satisfy the constraint. Justified-revision semantics assigns to any database a set (possibly empty) of its revisions. Each revision satisfies the constraints, and all deletions and additions of elements in a transition from initial database to the revision are derived from revision rules.

In this talk, we present an elegant embedding of revision programs into logic programs, which does not increase the size of a program. The embedding naturally led to extensions of the formalism of revision programming corresponding to existing extensions of logic programming. We also discuss annotated revision programs, which allow annotations like confidence factors, combined opinions of multiple experts, etc.

February 25nd 2002, 4:00pm, Ungar 506

Dr. Tomas Sander
InterTrust Technologies

Cryptographic Tools and Realistic Models for DRM

The modern digital world, in which computation and communication are (nearly) free and (almost) unlimited, poses critical new challenges for management of rights and information. In this talk, we will discuss some cryptographic and mathematical aspects of the concepts and mechanisms of to ensure security of and user privacy in Digital Rights Management (DRM) technology.

It is clear that cryptographic techniques can be useful in the design of a DRM system. However, they must be applied with care to the DRM setting. For example, unlike the situation in some other cryptographic scenarios, the typical end users of a DRM system may not be fully trusted by the designers of the system. Furthermore, security measures must be as unobtrusive as possible if the DRM system is going to be successfully deployed. Both perspectives lead to interesting requirements for cryptographic solutions. Besides describing the mathematical foundations for useful cryptographic techniques we will also illustrate these real-world constraints with a number of examples as they apply to the architecture of a DRM system.

Dr. Tomas Sander is a Member of the research team at the "Strategic Technologies and Architectural Research Laboratory (STAR Lab)" at InterTrust Technologies, Santa Clara, California. Prior to joining InterTrust in 1999, Tomas spent three years as a postdoctoral fellow at the International Computer Science Institute (ICSI) in Berkeley, California, for which he had received a fellowship from ICSI. Tomas received his doctoral degree in mathematics summa cum laude from the University of Dortmund, Germany in 1996. He has published and lectured scientific articles in both the practical and theoretical aspects of cryptology, computational mathematics, the theory of computing, electronic commerce and Digital Rights Management. He was the program chair of the ACM Workshop on Security and Privacy in Digital Rights Management (Nov. 2001, Philadelphia).

February 22nd 2002, 4:00pm, Ungar 506

Dr Holger H. Hoos
University of British Columbia

On the Behaviour of Stochastic Local Search Algorithms for SAT:
New Insights and Empirical Approaches

The propositional satisfiability problem (SAT) is one of the most prominent and conceptually simplest combinatorial problems in computing science. Over the past ten years, substantial progress has been made in the development of stochastic local search (SLS) algorithms for SAT; empirical methods played a key role in improving our ability to solve large and hard SAT instances from a broad range of domains, and increasingly help us to gain a better understanding for the behaviour of state-of-the-art algorithms. Recently, advanced empirical methods, in particular the RTD-based approach for analysing and characterising the behaviour of randomised algorithms for SAT, have been used to study some of the best-performing SAT solvers known todate.

In this talk, I will first outline the empirical approach I use for analysing and characterising the run-time behaviour of SLS algorithms for SAT and other problems. I will then present some of the most recent results I obtained using this methodology and present a surprisingly simple and general model for the behaviour of SLS algorithms for SAT that is based on these results and accounts for many observations that were previously not well understood. Based on these insights, I developed a new, self-tuning variant of Novelty+, one of the best-performing SLS algorithms for SAT. This new algorithm achieves performance levels close to the peak performance reached by Novelty+ using instance-specific, manually tuned parameter settings.

Finally, I will give a brief overview of various lines of work that demonstrate how many insights obtained for SAT can be used for improving the behaviour of SLS algorithms for other combinatorial problems, including the winner-determination problem in combinatorial auctions and a code design problem relevant to DNA computing and molecular tagging techniques.

For further information on the presenter, see http://www.cs.ubc.ca/~hoos`

February 20th 2002, 5:00pm, Ungar 411

Dr Dilip Sarkar
University of Miami

CAC Algorithms for QoS in Wireless Mobile Systems

Future broadband wireless access systems plan to integrate various classes of mobile terminals (MTs), each class with a different type of quality of service (QoS) requirement. When the load on a wireless network is high, the guarantee of QoS for each class of MTs is a challenging task. The measure of QoS is the probability of forced termination of a call that was allowed to access the network. Two previous handoff prioritization schemes - (i) prerequest scheme and (ii) guard channel scheme - decrease handoff failure (and hence forced termination). In this work, we compare and contrast both the schemes through extensive analysis and simulation; it is found that neither method can guarantee a desired level of QoS. We then propose a novel Call Admission Control (CAC) algorithm that can maintain any desired level of QoS, while successful call completion rate is very high. In the proposed algorithm, the new call arrival rate is estimated continuously, and when the estimated arrival rate is higher than a predetermined level, some new calls are blocked irrespective of the availability of channels. The objective of this pre-blocking of calls is to maintain the wireless network system's observed new call arrival rate at no more than a predetermined rate. We show that the proposed method can guarantee any desired level of QoS for profiled users.

This is the 4th in the Departmental Pizza Seminar series.

February 13th 2002, 5:00pm, Ungar 411

Erich Ruth (PhD candidate)
University of Miami

Estimating the Energy Diminishing Path of Proteins

Self-assembly is the ability to establish a complex, hierarchical structure from simpler, independent entities without external intervention. Amino acids, one of life's basic biological structures, can spontaneously assemble in solution and fold into a protein. The arrangement of amino acids determines a protein's folded shape which in turn dictates its specific function. The amino acids must unite in a precise sequential order to form a functioning protein, much like letters in a language must be properly sequenced to make words, sentences, and poetry. Understanding the self-assembly process of proteins would make it possible to design proteins that carry out a specific task by linking together the appropriate amino acid sequence(s).

Proteins assume a number of intermediate conformations before reaching a local or global energy minimum. In this paper, a linear programming-based algorithm is applied to previously developed energy formulations in an effort to calculate the energy diminishing path of a protein in an aqueous solution. Background on the structure of proteins, our empirical energy functions, and our linear programming, position-based algorithm is provided. We then develop the theory behind optimization of single-stranded proteins. We explain how we calculate the energy-diminishing path of a protein and compare our results with those of other software. A recommendation for future study is included.

It is an exciting time in the history of science. While the DNA contains the blueprint for the creation and evolution of life, protein's are life's machinery that perform the important tasks. Solving the protein-folding problem would open a world of new medical opportunities. This endeavor is challenging, contains a rich source of interesting problems and combines several academic fields, notably mathematics, computer science, and biochemistry.

This is the 3rd in the Departmental Pizza Seminar series.

February 11th 2002, 4:00pm, Ungar 426

Dr. Krishna Sivalingam
Washington State University

On the Routing, Wavelength and Time-Slot Assignment Problem in
Optical Wavelength Division (WDM) Wide Area Networks

Optical Wavelength Division Multiplexed (WDM) networks offer the potential of substantially higher bandwidths in the order of multi-gigabit to terabits per second. Wavelength routed WDM networks are an attractive candidate for the next generation Internet and beyond. In such networks, the physical topology of the network may be represented as a graph with nodes representing wavelength routers, interconnected by multi-wavelength links. Given a session request, the routing and wavelength assignment (RWA) problem is to calculate the shortest-path between two nodes, and also assign a set of wavelengths along this path. In a conventional wavelength routed network, an entire wavelength is assigned to a given session (or circuit). This can lead to lower channel utilization when the individual sessions do not need the entire channel bandwidth. This work considers a TDM-based approach to reduce this inefficiency.

In this network architecture, each individual wavelength is partitioned in the time-domain into fixed-length time-slots organized as a TDM frame. Multiple sessions are multiplexed on each wavelength by assigning a sub-set of the TDM slots to each session. Thus, given a session request with a specified bandwidth, the goal is to determine the route, wavelength and time-slot assignment (RWTA) that meets the given request. In this work, we present a family of RWTA algorithms and study the resulting blocking performance. For routing, we use shortest-path routing that uses link costs based on a least resistance weight function, which incorporates link load and number of hops to choose the path. For wavelength assignment, we employ the known least loaded wavelength selection algorithm; and for time-slot allocation, we present the least-loaded time-slot algorithm with three different variations. Simulation based analyses are used to compare the proposed RWTA algorithm to traditional wavelength-routed networks with and without wavelength conversion. The goal is to compare the benefits of TDM and wavelength conversion in improving performance. The results show that the TDM/WDM architecture provides substantial gains, especially for multi-fiber networks.

This work was supported in part by Cisco Systems, San Jose, CA.

Dr. Krishna Sivalingam is an Associate Professor in the School of Electrical Engineering and Computer Science, at Washington State University, Pullman, where he was an Assistant Professor from 1997 to 2000. From 1994 to 1997, he was an Assistant Professor at University of North Carolina Greensboro. During summer months of 1996 and 1997, he conducted research at Lucent Technologies' Bell Labs in Murray Hill, NJ, and at AT&T Labs in Whippany, NJ respectively. He also served as consultant to AT&T Labs during 1997. He received his Ph.D. and M.S. degrees in Computer Science in 1994 and 1990 respectively from State University of New York at Buffalo where he was a Presidential Fellow from 1988 to 1991. Prior, he received his B.E. degree in Computer Science and Engineering in 1988 from Anna University, Madras, India.

His research interests include wireless and mobile networks, optical WDM networks, and performance evaluation. He is co-recipient of the Best Paper Award at IEEE International Conference on Networks (ICON 2000) held at Singapore in Sep. 2000. He served as Lead Guest Editor for a special issue of the IEEE Journal on Selected Areas in Communications in optical WDM networks. Dr. Sivalingam has co-edited a book on optical WDM networks, published by Kluwer Academic Publishers in March 2000. He has served on committees of several international conferences, and is presently serving as Technical Program Co-Chair of SPIE/IEEE/ACM OptiComm conference to be held at Boston, MA in July 2002. His work has been supported by AFOSR, DoD Laboratory for Telecommunication Sciences, NSF, Cisco, Alcatel, Intel, Telcordia Technologies, and Washington Technology Center.

February 6th 2002, 5:00pm, Ungar 411

Dr. Christian Duncan
University of Miami

Optimal Constrained Graph Exploration
From firemen to treasure hunters

In this talk, we address the problem of constrained exploration of an unknown graph G=(V,E) from a given start node s with either a tethered robot or a robot with a fuel tank of limited capacity, the former being a tighter constraint. In both variations of the problem, the robot can move along only the edges of the graph, i.e, it cannot jump between non-adjacent vertices. In the tethered robot case, if the tether (rope) has length l, then the robot must remain within distance l from the start node s. In the second variation, a fuel tank of limited capacity C forces the robot to return to s after traversing C edges. The efficiency of algorithms for both variations of the problem is measured by the number of edges traversed during the exploration. We present an algorithm for a tethered robot which explores the graph in q(|E|) edge traversals. The problem of exploration using a robot with a limited fuel tank capacity can be solved with a simple reduction from the tethered robot case and also yields a q(|E|) algorithm. This improves on the previous best known bound of O(|E| + |V|log2|V|). Since the lower bound for the graph exploration problems is |E|, our algorithm is optimal, thus answering the open problem of Awerbuch, Betke, Rivest, and Singh.

We also extend our results to a modified version of the shortest path problem called the treasure hunting problem. Here a robot wishes to quickly discover some treasure from an unknown graph. We conclude our talk with some interesting open problems.

This is the 2nd in the Departmental Pizza Seminar series.

January 30th 2002, 5:00pm, Ungar 411

Dr. Geoff Sutcliffe
University of Miami

The Design and Implementation of a
Compositional Competition-Cooperation Parallel
ATP System

Key concerns in the development of more powerful ATP systems are to provide breadth of coverage - an ability to solve a large range of problems, and to provide greater depth of coverage - an ability to solve more difficult problems, within the same resource limits. This work describes the design and implementation of CSSCPA, a compositional competition-cooperation parallel ATP System. CSSCPA combines existing high performance ATP systems in a framework that allows them to work independently, but also allows communication of intermediate results. The performance data shows that CSSPCA has high breadth and depth of coverage.

This is the 1st in the Departmental Pizza Seminar series.

January 14th 2002, 4:00pm, Ungar 426

Dr. Simon Colton
University of Edinburgh

Automated Discovery in Pure Mathematics

Various computational techniques have led to discoveries in pure mathematics. These techniques include computer algebra, automated theorem provers, constraint solvers and automated theory formation programs. I will give a brief survey of some discoveries made in this way, then concentrate on our particular approach: the HR automated theory formation program. I will give some details of how HR works, and then discuss some of the mathematical results it has discovered in various domains. I will conclude with an exploration of refactorable numbers, which are such that the number of divisors is itself a divisor.

December 14th 2001, 11:00am, Ungar 426

Dr. Tapas K Nayak
Microsoft, Seattle

Perspectives and Issues in Text-indexing

Until recently database systems were totally focused to organization and retrieval of structured data. Unstructured and semi-structured data were being managed by independent software packages built on top of file system stores or some other stores not integrated tightly with database systems. In recent times, however, with explosion of information people have started to realize importance of a tighter integration of such data with database store for the sake of inter-operability, reliability, scalability, maintainability and efficiency. Building full-text indexes as integral part of a database system is a first attempt towards that goal. Full-text indexes, inherently, come in with problems not seen usually in database applications before. Problems are mostly around the issue of scale. The talk will highlight such issues and how they impact design strategies and how some of them are addressed in present versions of Microsoft SQL Server.

Dr. Tapas K Nayak did his PhD in Computer Science from Indian Institute of Technology, Kharagpur, India, in 1992. He is presently with Microsoft at Seattle and is involved in architecting future version of full-text indexing in Microsoft SQL Server. He had earlier been a faculty at Indian Institute of Technology, Kanpur, India. His current interests lie in databases, data structures, very large system design, and computer languages.

December 11th 2001, 3:00pm, Ungar 506

Dr. Bruce Litow
James Cook University

Linear Diophantine Equations, Semilinear Sets and Rational Series

In this talk we survey the close connection between the solution set of a finite system of linear Diophantine equations and semilinear sets. Linear sets generalize arithmetic progressions to several dimensions. The closure under union of the linear sets yields the semilinear sets, which by a kind of miracle are closed under all Boolean operations.

We give a high-level, conceptually transparent algorithm that reveals both the structure of the solution set of a system of linear Diophantine equations and its computational complexity. Along the way we also point out an important connection between the set of support of multivariate rational series and general Diophantine polynomials.

Dr. Bruce Litow is a senior lecturer in the School of Information Technology at James Cook University. His principal research interest is the computational complexity of arithmetic and algebraic problems.

May 14th 2001, 4:00pm, Ungar 506

Stephan Schulz
Technische Universität München

Learning Search Control Knowledge for Equational Theorem Proving

Automated theorem proving (ATP) is an area of growing importance in many industrial and scientific applications. In particular, ATP systems are increasingly being used to formalize mathematical theories and results, or to verify the design of hard- and software products. Most of the current high-performance ATP systems are based on an equational paradigm.

However, while such ATP systems are extremely versatile and powerful tools, this power comes at a price. ATP systems for most interesting logics have to search for proofs in an infinite search space. This search is typically controlled by heuristic evaluation functions, which are used to select the "best" alternative at the important choice points during the search. The performance of nearly every fully automatic ATP system depends critically on the quality of these evaluation functions.

In this talk, I will present an approach to learn good evaluation heuristics from many previous proof experiences. Search decisions are represented by patterns of clauses, and evaluations for these patterns are learnt by analysis of the use of matching clauses in previous proof searches.

The approach was successfully implemented and tested in the superposition-based prover E/TSM. The talk concludes with a live demonstration of the proof system.