2017-2018

Fall 2017 Abstracts

Graduate Orientation II: Presentation, Poster Session, & PIZZA!


MANDATORY FOR NEW GRAD STUDENTS and counts towards requirement for CSCE 681 students.

4:10-6:00 p.m., Wednesday August 30, 2017
Room 124, Bright Building

Abstract

  • 4:10-5:10 p.m. - Presentation
  • 5:10-6:00 p.m. - Pizza & Current Student Poster Session - new students can meet current grads and learn about ongoing research projects.

Graduate Orientation I: Overview of Department Resources & Contacts, Honor Code, and Student Organizations


MANDATORY FOR NEW GRAD STUDENTS (but not other CSCE 681 students)

4:10-6:00 p.m., Monday September 4, 2017
Room 124, Bright Building

Abstract

This meeting will concentrate on the essentials the students will need to settle in. It will include an introduction to departmental administration (staff who's who, payroll, mailboxes, phones), the computing resources (computer use/accounts, printer quotas, lab access/tours), the academic advising staff and resources, the TAMU honor system, TAMU Libraries, Graduate Teaching Academy, Student Engineers' Council and relevant student organizations (CSEGSAAWICSTACS (TAMU ACM and IEEE student chapter), UPE and TAGD).


Systems for Clinical Outcomes Predictions

Bobak Mortazavi
Assistant Professor
Texas A&M University

4:10pm Wednesday, September 6, 2017
HRBB 124

Abstract

The design of personal medical embedded systems for user-centric health monitoring involves an understanding of platform development for data collection, applications of machine learning for processing vast quantities of varying data, and an understanding of the underlying clinical questions these systems are trying to address. The interdisciplinary nature of these tasks requires an understanding of the clinical issues being addressed, and then developing specific systems and algorithms to address these, along with the unique challenges posed by each individual application. This talk focuses on an understanding of clinical data, the challenges posed by implementing machine learning techniques, understanding the differences between methods used in clinical outcomes predictions and those available to computer scientists, and then examines several open-ended case studies that have the potential for both algorithmic and embedded systems improvements.

Biography

Jack Bobak Mortazavi is an Assistant Professor in Computer Science and Engineering at Texas A&M. After receiving his bachelors degree from the University of California Berkeley, he earned his Ph.D. in Computer Science from the University of California Los Angeles, where he focused on the development of embedded systems for the Wireless Health Institute. Most recently, he was a postdoctoral associate, and then instructor, in the department of internal medicine, section of cardiology, at the Yale School of Medicine. He has recently focused on clinical research challenges in predictive models and comparative effectiveness techniques, in order to better address the challenges of personalized health monitoring, in order to develop personalized remote systems for clinical outcomes.

Faculty Contact: Dr. Dilma Da Silva


Bio-behavioral signals and systems: From signal representations to novel health applications

Theodora Chaspari
Assistant Professor
Texas A&M University

4:10pm Wednesday, September 13, 2017
Room 124 HRBB

Abstract

Bio-behavioral signal processing and systems modeling enable an integrated computational approach to the study of human behavior and human physical and mental well-being through overt behavioral signals information and covert biomarkers. Recent converging advances in sensing and computing, including wearable technologies, allow the unobtrusive long-term tracking of individuals yielding rich multimodal signal measurements from real-life. In this talk, we will present the development of data-scientific and context-rich bio-behavioral approaches for analyzing, quantifying, and interpreting these bio-behavioral signals. The first part of the talk will describe a novel knowledge-driven signal representation framework able to efficiently handle the large volume of acquired data and the noisy signal measurements. Our approach involves the use of sparse approximation techniques and the design of signal-specific dictionaries learned through Bayesian methods, outperforming previously proposed models in terms of signal reconstruction and information retrieval criteria. The second part will focus on translating the derived signal representations into novel intuitive quantitative measures analyzed with probabilistic and statistical models in relation to external factors of observable behavior. This work has found applications in Autism intervention for detecting beneficial regulation mechanisms during child-therapist interactions, as well as in the family studies domain for identifying instances of emotional escalation and interpersonal conflict. The final part of the talk will discuss how the results from this analysis can be employed toward designing human-assistive personalized bio-feedback systems able to promote healthy routines, increase emotional wellness and awareness, and revolutionize clinical assessment and intervention.

Biography

Theodora Chaspari is an Assistant Professor at the Computer Science & Engineering Department in Texas A&M University. She has received the diploma (2010) in Electrical and Computer Engineering from the National Technical University of Athens, Greece and the Master of Science (2012) and Ph.D. (2017) in Electrical Engineering from the University of Southern California. Since 2010 she is working as a Research Assistant at the Signal Analysis and Interpretation Laboratory at USC. She has also been a Lab Associate Intern at Disney Research (summer 2015). Dr Chaspari’s research interests lie in the areas of biomedical signal processing, human-computer interaction, behavioral signal processing, data science, and machine learning. She is a recipient of the USC Annenberg Graduate Fellowship, USC Women in Science and Engineering Merit Fellowship, and the IEEE Signal Processing Society Travel Grant.

Faculty Contact: Dr. Dilma Da Silva


Interactive Modeling Techniques for Trees and Mechanisms

Zhigang Deng
Professor
University of Houston

4:10pm Wednesday, September 20, 2017
Room 124 HRBB

Abstract

In recent years, how to effectively create high-quality 3D models for complex objects including trees and mechanisms have attracted a lot of attention. In this talk, I will present two recent works in this direction. The first work is an effective modeling technique to generate a set of morphologically diverse and inspiring virtual trees through hierarchical topology-preserving blending, aiming to facilitate designers’ creativity production. On top of that, I will further describe a morphing technique to generate high quality visual effects between topologically varying trees while preserving the topological consistency and botanical meanings of any in-between shapes as natural trees. The second work is a novel interactive system for mechanism modeling from multi-view images. Its key feature is that the generated 3D mechanism models contain not only geometric shapes but also internal motion structures: they can be directly animated through kinematic simulation.

Biography

Dr. Zhigang Deng is a (Full) Professor of Computer Science at University of Houston (UH). He is also the Director of Graduate Studies at UH Computer Science Department and the Director of the UH Computer Graphics and Interactive Media (CGIM) Lab. He earned Ph.D. in Computer Science at the University of Southern California in 2006. He also completed B.S. degree in Mathematics from Xiamen University (China), and M.S. in Computer Science from Peking University (China). Over the years, He has worked or consulted at the Founder Research and Development Center (China), AT&T Shannon Research Lab, and Qualcomm Research Center. His current research interests are in the broad, interdisciplinary areas of graphics/animation, human computer interaction, virtual human modeling & animation, affective computing, and humanoid robots. He is the recipient of many awards including CASA Best Paper Award (2017), ACM ICMI Ten Year Technical Impact Award (2014), UH Teaching Excellence Award (2013), NSFC Overseas and Hong Kong/Macau Young Scholars Collaborative Research Award (2013), ICRA Best Medical Robotics Paper Award Runner-up (2012), and Google Faculty Research Award (2010). Besides the CASA 2014 conference general co-chair and SCA 2015 conference general co-chair, he currently serves as an Associate Editor of several journals including Computer Graphics Forum, and Computer Animation and Virtual Worlds Journal. His research has been funded by NSF, NIH, NASA, DOD, Texas NHARP, and various industry sources (Google, Nokia, NVidia, etc). More information can be found at his webpage, http://graphics.cs.uh.edu/zdeng

Faculty Contact: Dr. Shinjiro Sueda


Exploiting Low-Quality Visual Data using Deep Networks

Zhangyang (Atlas) Wang
Assistant Professor
Texas A&M Univeristy

4:10pm Monday, September 25, 2017
Room 124 HRBB

Abstract

While many sophisticated models are developed for visual information processing, very few pay attention to their usability in the presence of data quality degradations. Most successful models are trained and evaluated on high quality visual datasets. On the other hand, the data source often cannot be assured of high quality in practical scenarios. For example, video surveillance systems have to rely on cameras of very limited definitions, due to the prohibitive costs of installing high-definition cameras all around, leading to the practical need to recognize objects reliably from very low resolution images. Other quality factors, such as occlusion, motion blur, missing data and bad weather conditions, are also ubiquitous in the wild. The seminar will present a comprehensive and in-depth review, on the recent advances in the robust sensing, processing and understanding of low-quality visual data, using deep learning methods. I will mainly show how the image/video restoration and the visual recognition could be jointly optimized as one pipeline. Such an end-to-end optimization consistently achieves the superior performance over the traditional multi-stage pipelines. I will also demonstrate how our proposed approach largely improves a number of real-world applications.

Biography

Dr. Zhangyang (Atlas) Wang is an Assistant Professor of Computer Science and Engineering (CSE), at the Texas A&M University (TAMU). During 2012-2016, he was a Ph.D. student in the Electrical and Computer Engineering (ECE) Department, at the University of Illinois at Urbana-Champaign (UIUC), working with Professor Thomas S. Huang. Prior to that, he obtained the B.E. degree at the University of Science and Technology of China (USTC), in 2012. Dr. Wang's research has been addressing machine learning, computer vision and multimedia signal processing problems using advanced feature learning and optimization techniques. He has co-authored around 40 papers, and published several books and chapters. He has been granted 3 patents, and has received over 15 research awards and scholarships. His research has been covered by worldwide media, such as BBC, Fortune, International Business Times, UIUC news and alumni magazine. More could be found at: http://www.atlaswang.com

Faculty Contact: Dilma Da Silva


SuiteSparse:GraphBLAS: graph algorithms via sparse matrix operations on semirings

Tim Davis
Professor
Texas A&M University

4:10pm Monday, October 2, 2017
Room 124 HRBB

Abstract

SuiteSparse:GraphBLAS is an full implementation of the GraphBLAS standard, which defines a set of sparse matrix operations on an extended algebra of semirings using an almost unlimited variety of operators and types.  When applied to sparse adjacency matrices, these algebraic operations are equivalent to computations on graphs.  GraphBLAS provides a powerful and expressive framework for creating graph algorithms based on the elegant mathematics of sparse matrix operations on a semiring.

Performance of SuiteSparse:GraphBLAS is either on par with the corresponding operations in MATLAB, or faster.  Submatrix assignment is particularly efficient.  In one example, C(I,J)=A for a matrix C of size 3 million-by- 3 million with 14 million nonzeros, and a matrix A of size 5500-by-7000 with 38500 nonzeros, takes 82 seconds in MATLAB but only 0.74 seconds in SuiteSparse:GraphBLAS.  This result includes finalizing the computation and returning the result to MATLAB as a valid sparse matrix.  SuiteSparse:GraphBLAS also includes a non-blocking mode, so that a sequence of submatrix assignments can be still more efficient.

Biography

Tim Davis is a Professor in the Computer Science and Engineering Department at Texas A&M University.  His primary scholarly contribution is the creation of widely-used sparse matrix algorithms and software.  As an NVIDIA Academic Partner, he is creating a new suite of highly-parallel sparse direct methods that can exploit the high computational throughput of recent GPUs. He was elected in 2013 as a SIAM Fellow, in 2014 as an ACM Fellow, and in 2016 as an IEEE Fellow.  He serves as an associate editor for ACM Transactions on Mathematical Software, and the SIAM Journal on Scientific Computing.  Tim is a Master Consultant to The MathWorks, and the primary author of x=A\b in MATLAB when A is sparse.

Faculty Contact: Dr. Lawrence Rauchwerger


A Sub-linear Time Algorithm for Pattern Matching in Big Data

Krishna Narayanan
Eric D. Rubin '06 Professor
Texas A&M University

4:10pm Monday, October 9, 2017
Room 124 HRBB

Abstract

We are witnessing an unprecedented growth in the amount of data that is being collected and made available for data mining. While the availability of large-scale datasets presents exciting opportunities for advancing sciences, healthcare, understanding of human behavior etc., mining the data set for useful information becomes a computationally challenging task. We are in an era where the volume of data is growing faster than the rate at which available computing power is growing, thereby creating a dire need for computationally efficient algorithms for data mining. From an algorithmic complexity standpoint, we are transitioning from a mindset where algorithms with linear complexity in the size of the data set were considered efficient to an era when algorithms with linear complexity have become infeasible owing to the large size of the datasets. This necessitates the creation of algorithms with sub-linear time complexity tailored to big data.

One of the most fundamental data analytics tasks is that of querying a data set to see if a particular pattern of symbols appears in the data set either exactly or approximately. We assume that sketches of the original signal can be computed offline and stored. After providing a brief overview of existing algorithms for fast substring matching, I will describe our algorithm for substring matching which leverages the sparse Fourier transform computation-based approach introduced by Pawar and Ramchandran.  We show that our algorithm can find matches with high probability (asymptotically in the size of the dataset and the query) with sub-linear time complexity.

Potential applications of this work include text matching, audio/image matching, DNA matching in genomics, metabolomics, radio astronomy, searching for signatures of events within large databases, detecting viruses within binary executable files. I am actively looking for collaborators who can use fast pattern matching in their area of expertise.

Biography

Krishna Narayanan is the Eric D. Rubin'06 professor in the ECEN department at Texas A&M University. His research interests are in coding theory, information theory, and signal processing with applications to wireless communications, data storage, and data science. On the teaching side, he is excited by the use of technological tools to personalize the learning experience of students. He is a Fellow of the IEEE. He currently serves as an associate editor for the IEEE Transactions on Information Theory and also serves on the board of governors for the IEEE Information Theory society. When he is not matching patterns at work, he (mostly unsuccessfully) tries to identify patterns when listening to Indian classical music. He is also a self-proclaimed expert in analyzing cricket matches. 

Faculty Contact: Dr. Anxiao (Andrew) Jiang


Distributed Algorithmic Foundations of Large-scale Data Computation

Gopal Pandurangan
Professor
University of Houston

4:10pm Monday, October 16, 2017
Room 124 HRBB

Abstract

Motivated by the emergence of distributed ``Big Data" computing, we develop a theory of distributed computing for large-scale data. Our computation model is a distributed message-passing model called the k-machine model,  where we have k machines that jointly perform a computation on some input data of size n (typically, n is much larger than k). The input is assumed to be  partitioned among the k machines, which is a common situation in many real world systems. In particular, we focus on computation on graphs, and we present several complexity results --- both lower and upper bounds ---  for various fundamental graph problems such as verifying graph connectivity, constructing a minimum spanning tree, computing the PageRank, and enumerating triangles.  Our model provides a unified framework to design and analyze distributed algorithms for large-scale problems as well as quantity the fundamental limitations of distributively solving problems where the input is partitioned across a distributed system. 

Biography

Gopal Pandurangan (http://www.cs.uh.edu/~gopal) is a Professor in the Department of Computer Science at the University of Houston. He received his Ph.D.  in Computer Science from Brown University in 2002. He has held faculty and visiting positions at Nanyang Technological University in Singapore, Brown University, Purdue University, and Rutgers University. His research interests are in theory and algorithms, distributed computing, networks,  large-scale data, and computational biology. He has published over 100 refereed papers in these areas. His work has appeared in JACM, SICOMP, ACM TALG, STOC, FOCS, SODA, PODC, SPAA, INFOCOM, and RECOMB. His research has been supported by research grants from the US National Science Foundation, US-Israeli Binational Science Foundation, and the Singapore Ministry of Education.

Faculty Contact: Dr. Jennifer Welch


Detecting and Identifying Sign Language Content on Video Sharing Sites

Frank Shipman
Professor
Texas A&M University

4:10pm Monday, October 23, 2017
Room 124 HRBB

Abstract

Video sharing websites are used by members of the deaf and hard of hearing community to exchange signed content.  Unfortunately, these services lack the ability to search and locate untagged or unlabeled sign language content. As a result, members of this community rely on ad-hoc mechanisms to pass around pointers to internet-based recordings, such as email, blogs, etc.  To remedy this situation, we are developing techniques that automatically detect sign language video and that can distinguish between different sign languages. This talk describes our initial video-analysis techniques to detect sign language content, three optimization strategies to reduce the computational costs associated with such detection, and techniques to distinguish between different sign languages.

Biography

Frank Shipman is a Professor in the Department of Computer Science and Engineering at Texas A&M University. His research interests include topics in computer-supported cooperative work, multimedia, computers and education, and intelligent user interfaces and has published over 150 peer-reviewed papers on these topics. His current projects explore (1) issues surrounding social media ownership, (2) the potential of prediction games to promote data analysis skills, and (3) access to sign language video for the deaf and hard-of-hearing.

Faculty Contact: TBA


Internet privacy: Towards more transparency

Balachander Krishnamurthy
Lead Inventive Scientist
AT&T Labs – Research

4:10pm Monday, October 30, 2017
Room 124 HRBB

Abstract

Internet privacy has become a hot topic recently with the radical growth of Online Social Networks (OSN) and attendant publicity about various leakages. For the last several years we have been examining aggregation of user's information by a steadily decreasing number of entities as unrelated Web sites are browsed. I will present results from several studies on leakage of personally identifiable information (PII) via Online Social Networks and popular non-OSN sites. Linkage of information gleaned from different sources presents a challenging problem to technologists, privacy advocates, government agencies, and the multi-billion-dollar online advertising industry. Economics might hold the key in increasing transparency of the largely hidden exchange of data in return for access of so-called free services. I will also talk briefly about transient online social networks and doing privacy research at scale.

Biography

Balachander Krishnamurthy is a lead inventive scientist at AT&T Labs – Research. His focus of research is in the areas of Internet privacy, transparency and fairness in ML algorithms, and Internet measurements. He has authored and edited ten books, published over one hundred technical papers, holds seventy-five patents, and has given invited talks in thirty-five countries.

He co-founded the successful ACM Internet Measurement Conference in 2000 and in 2013 the Conference on Online Social Networks and is involved in the Data Transparency Lab efforts to fund privacy research. He has been on the thesis committee of several Ph.D. students, collaborated with over eighty researchers worldwide, and given tutorials at several industrial sites and conferences.

His book "Internet Measurements: Infrastructure, Traffic and Applications" (525pp, Wiley, with Mark Crovella) is the first book focusing on Internet Measurement. His previous book with Jen Rexford, 'Web Protocols and Practice: HTTP/1.1, Networking Protocols, Caching, and Traffic Measurement' (672 pp, Addison-Wesley), is the first in-depth book on the technology underlying the World Wide Web, and has been translated into Portuguese, Japanese, Russian, and Chinese. Bala is homepageless and not on any OSN but many of his papers can be found at http://www.research.att.com/~bala/papers.

Faculty Contact: Dr. Nick Duffield


TBA

Chao Tian
Associate Professor
Texas A&M University

4:10pm Wednesday, November 1, 2017
Room 124 HRBB

Abstract

TBA

Biography

TBA

Faculty Contact: Dr. Anxiao (Andrew) Jiang


TBA

J. Maurice Rojas
Professor of Mathematics and Computer Science and Engineering
Texas A&M University

4:10pm Monday, November 6, 2017
Room 124 HRBB

Abstract

TBA

Biography

TBA

Faculty Contact: Dr. Nancy M. Amato


Balancing Naturalness, Convenience, and Comfort for Interaction Technique in Virtual Reality

Eric Ragan
Assistant Professor
Texas A&M University

4:10pm Monday, November 13, 2017
Room 124 HRBB

Abstract

For virtual reality (VR), the goal is often to achieve a realistic simulation that closely matches the experiences of the real world. It would be ideal if users could freely walk around and use their physical hands to directly interact with the virtual environment. While this can be achieved via interaction with tracking technology, practical limitations can make it difficult to simulate a variety of situations with a single system. For example, virtual environments are often much larger than the available tracked physical space, and virtual worlds can have more ways to interact than a typical physical space supports. A related issue with realistic interaction is that large amounts of physical movement are often not preferred for comfortable and convenient use of technology. Our research investigates interaction techniques that balance the level of realism with level of convenience for practical real-world uses of VR. We study semi-natural methods for navigation and view control that can work for seated use of virtual reality with HMDs when physically turning all the way around is not ideal, such as when sitting on a couch or at a desk. We also explore the use of perceptual illusions and haptics to allow direct hand interaction through physical props. This talk will provide an overview of the technical and practical considerations important for the design of convenient 3D interaction techniques, and it will present the results of empirical studies of how different techniques affect users’ spatial orientation, sickness, and experiences in VR.

Biography

Dr. Eric Ragan is an Assistant Professor in the Department of Visualization at Texas A&M University, and he is a joint faculty member in the Department of Computer Science and Engineering. He directs the Interactive Data and Immersive Environments (INDIE) Lab. His research interests include human-computer interaction, information visualization, visual analytics, virtual reality, and 3D interaction techniques. He previously worked as a visual analytics research scientist at Oak Ridge National Laboratory, where he studied visualization designs that enable monitoring and analysis of streaming data. Current research topics include the visualization of analytic provenance, understandable visual interfaces for machine learning systems, and the natural interaction techniques for immersive virtual environments. Dr. Ragan received his Ph.D. in computer science from Virginia Tech. Contact him at eragan@tamu.edu.

Faculty Contact: Dr. John Keyser