4:00-5:00 p.m., Monday August 29, 2005
Room 124, Bright Building
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, and relevant student organizations (CSGSA, AWICS, TACS (TAMU ACM and IEEE student chapter), and UPE ).
MANDATORY FOR NEW GRAD STUDENTS (but not other CSPC 681 students)
4:00-6:00 p.m., Wednesday August 31, 2005
Room 124, Bright Building
MANDATORY FOR NEW GRAD STUDENTS and CPSC 681 STUDENTS
Dr. Tiffani Williams, Assistant Professor of Computer Science, Texas A&M University
4:10 p.m., Wednesday September 7, 2005
Room 124, Bright Building
Phylogenetic (or evolutionary) trees model the evolution of biological species or genes from a common ancestor. Scientists are interested in evolutionary trees for the usual reasons of scientific curiosity. However, phylogenetic analysis is not just an academic exercise. Phylogenies are the organizing principle for most biological knowledge. As such, they are a crucial tool in identifying emerging diseases, predicting disease outbreaks, and protecting ecosystems from invasive species. The greatest impact of phylogenetics will be reconstructing the Tree of Life, the evolutionary history of all-known organisms.
In this talk, I will focus on obtaining phylogenetic analyses of large datasets under NP-hard optimization criteria (i.e., Maximum Parsimony and Maximum Likelihood) quickly and accurately. Our current techniques are able to analyze datasets containing thousands of taxa much faster than traditional methods. I will also discuss promising work with parallel computing, using the coordination language Linda, to integrate a multitude of sequential phylogenetic algorithms for even greater performance.
Tiffani L. Williams is an Assistant Professor in the Department of Computer Science at Texas A&M University. During the 2004-2005 academic year, she was the Edward, Frances, and Shirley B. Daniels Fellow at the Radcliffe Institute of Advanced Study at Harvard University. She earned her B.S. in computer science from Marquette University and Ph.D. in computer science from the University of Central Florida. Afterward, she was a postdoctoral fellow at the University of New Mexico. Her honors include a Radcliffe Institute Fellowship, an Alfred P. Sloan Foundation Postdoctoral Fellowship, and a McKnight Doctoral Fellowship. Her research interests are in the areas of bioinformatics and high-performance computing.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Ismael Lopez, CIATEQ (Center for Advanced Technologies), Mexico
4:10 p.m., Wednesday September 14, 2005
Room 124, Bright Building
Robot performance during manipulative tasks has traditionally depended on simple sensing systems and the robot manufacturers programming language. However, this restricts the use of robots in complex manufacturing operations. An alternative to robot programming is the creation of self-adaptive robots based on Artificial Intelligence techniques.
The research presented in this talk shows how industrial robots can operate autonomously in unstructured environments. This is achieved by providing the robot with a primitive knowledge base (PKB) of the environment in terms of visual and tactile information. This knowledge is gradually enhanced on-line based on the contact force information acquired during operations. The robot is able to recognize and learn different assembly component parts and also able to grasp and to perform autonomously the assembly operation.
When assembling the components, the robot resembles a blindfold person performing the same task since no precise information is provided about the localisation of the xed assembly component. The design of a novel neural network controller (NNC) based on the Fuzzy ARTMAP network and its implementation results on industrial robots are presented.
Finally, ongoing work looking towards the creation of a Multimodal Architecture (M2ARTMAP) to integrate other sensorial information (e.g. odour and taste), to reinforce the prediction capability is presented together with simulated results.
Dr. Lopez-Juarez obtained a BEng in Mechanical and Electrical Engineering from The National Autonomous University of Mexico (UNAM) in 1991. He lectured at the Faculty of Engineering during 1991-1994 and also worked as Research Assistant in the Department of Computing Systems and Automation at The Institute of Research on Applied Mathematics and Systems (IIMAS-UNAM). He obtained an MSc in Instrument Design and Application at UMIST and a PhD in Intelligent Robotics at The Nottingham Trent University in collaboration with Rolls Royce & Associates during 1996 and 2000, respectively.
Dr. Lopez-Juarez has published over forty papers in referred international journals and proceedings in the area of Instrumentation, Self-adaptive Industrial Robots and Neural Networks. He has also served as reviewer of International Conferences and Journals. He is the founder and leader of the Mechatronics and Intelligent Manufacturing Systems Research Group (MIMSRG) at CIATEQ. He is responsible for several industrially and government sponsored projects in the field of flexible manufacturing as well as two projects with Texas A&M University and the German Academic Interchange Service (DAAD) in the field of instrumentation and flexible manufacturing. He is a founder member of the Mexican Society on Mechatronics (AMM), member of the Mexican Society of Robotics (AmROB), member of the Mexican Society for Computer Science (SMCC), member of the IEEE-Robotics and Automation Society and also member of the IEEE-Computational Intelligence Society.
CPSC 681 Faculty Contact: Ricardo Gutierrez-Osuna (rgutier [at] cs.tamu.edu)
Dr. Shay Kutten, Professor of Industrial Engineering, The Technion, Israel
4:10 p.m., Monday September 19, 2005
Room 124, Bright Building
Traditional fault tolerance methods are global in nature. Some involve resetting all the network nodes. Others use time-outs that imply waiting to the slowest and furthest node, etc.
Local detection of faults is based on the following observation: even computations that cannot be performed locally, can be verified locally. Hence, it is possible to verify locally (cooperating only with neighbors) global correctness predicates. If the predicate does not hold, then measures for correctness can be invoked.
The cost of correction can also be, sometimes, local to the faults, in the sense that when the extent of the faults is lower, the cost of the recovery is lower too.
The talk will mention results from several papers by several authors, including a very recent paper.
Shay Kutten received his Master and his PhD in Computer Science from the Technion, Israel, in 1984 and 1987 respectively. From 1987 to 1996 he was with IBM T.J. Watson Research Center, as a post doctoral fellow, as a project leader, as the manager of the Network Architecture and Algorithms group, and as a Research Staff Member.
At the Technion he is now the coordinator of undergraduate studies of the faculty of Industrial Engineering and Management. He won the Taub Award for excellence in research, and the Mitchner Award for research on Quality Sciences and Quality Management. He is an area editor (for security, reliability, and availability) of the ACM's journal on Selected Topics in Mobile Networks and Applications (MONET) and the Elsevier journal Computer Networks. He is a senior member of the IEEE, and a member of ACM-SIGACT.
CPSC 681 Faculty Contact: Jennifer Welch (welch [at] cs.tamu.edu)
Kathleen Dollard, Gendotnet
4:10 p.m., Wednesday September 21, 2005
Room 124, Bright Building
The new generics feature of .NET 2005 has gotten a lot of press but what's it actually good for? Walk through the syntax of generics and basic rules regarding their use, and then explore how to integrate them into your applications and architectures - both the new generic classes in the framework and your own generic classes. Learn how to design generic classes at the core of your infrastructure, and how to avoid dead ends where you think generics will do something that they won't. The talk will compare, contrast, and combine generics and OOP models such as inheritance and interfaces to introduce the new architectural patterns available in .NET 2005, as well as clarifying details such as how and when to use the new generic collections and nullable classes.
Kathleen Dollard is a nationally recognized author, trainer, and speaker. Microsoft has honored her with its "Most Valuable Professional" award every year since 1998. INETA included her in its prestigious speaker's bureau. She's written numerous articles and "Code Generation in Microsoft .NET" (Apress, 2004). She also codes almost every day as a consultant or for her own company GenDotNet. Her passion is helping programmers be smarter in how they develop. She's currently working on full life cycle improvements, such as better debugging and capturing business intent in metadata and test definitions. She received her Masters of Science in Chemistry at Texas A&M in 1983. She moved full time from chemistry to computers at the dawn of the PC revolution while working in industry in the 1980's, and has been an independent consultant since 1989.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Anxiao (Andrew) Jiang, Assistant Professor of Computer Science, Texas A&M University
4:10 p.m., Monday September 26, 2005
Room 124, Bright Building
One important routing method for large ad hoc wireless networks is geographic routing. It uses the locations of routers and a packet's destination to make packet-forwarding decisions. Geographic routing is efficient and scalable, however the location information is often expensive or impossible to obtain. Because of that, routing algorithms that generate 'virtual coordinates' of routers based on known information --- e.g., connectivity --- have been developed, thus avoiding the usage of true location information. In this talk, I will overview the above location-based and location-free routing algorithms, and then present a new and different location-free routing algorithm --- MAP --- that uses the medial axis of the network to name routers and make routing decisions.
I will show how location-based routing algorithms use planar spanning subgraphs to guarantee delivery of messages. However, for such techniques to work, the network needs to adhere to the very restrictive unit-disk graph model. Also, they create unbalanced load for networks deployed in environments with complex topologies. Location-free routing algorithms based on virtual coordinates often do not guarantee delivery. MAP, on the other side, uses medial axis --- an abstraction of the environment topology --- to enable an efficient and load-balanced routing strategy that is not restricted to the unit-disk graph model.
Anxiao (Andrew) Jiang is an assistant professor in the Computer Science Department of Texas A&M University. He received his Ph.D. and M.S. degrees in electrical engineering from the California Institute of Technology in 2004 and 2000, respectively. He received his B.S. degree in electronic engineering from Tsinghua University, Beijing, China in 1999. His research interests include algorithms, wireless ad hoc communication and sensor networks, file storage and retrieval, and distributed systems.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Ioannis Pavlidis, Associate Professor, University of Houston
4:10 p.m., Monday October 3, 2005
Room 124, Bright Building
People in the U.S. spend increasing amounts of their work and leisure time in front of computers. The term "human computer interaction" suggests a two-way exchange, with each participant aware of the other and responding appropriately. In reality, computers may appear frequently rude, frustrating, indifferent, and generally unpleasant. Much of this can be attributed to the fact that current computers are almost completely unaware of the actual state of the human user. The research community has proposed methods for computers to understand and respond to the computer user's feelings with mixed results. Emphasis has been placed on stress and frustration - feelings often generated by the use of the computer itself. There is no previous work, however, with regard to monitoring the actual health of the user during computer use. People check their health status rarely or when they are symptomatic. A typical health check features measurement of vital signs (e.g., pulse and respiration) at a doctor's office. The value of this check is limited by its isolated nature. This is especially true for chronic ailments like heartbeat irregularities, headaches, or anxiety disorders, which manifest themselves intermittently for short intervals in a stochastic manner. In this research effort we incorporate physiologic monitoring in the human-computer interface (HCI). The sensing element is a thermal imaging camera that is employed as a computer peripheral. Through bioheat modeling of facial imagery almost the full range of vital signs can be extracted, including blood flow, cardiac pulse, and breathing rate. This nearly continuous physiological information can then be used to draw inferences about a variety of health symptoms. Our research aims to realize the notion of desktop health monitoring and create truly collaborative interactions in which human and machines are both observing and responding.
Dr. Pavlidis holds a Ph.D. and a M.S. degree in Computer Science from the University of Minnesota, a M.S. degree in Robotics from the Imperial College of the University of London, and a B.S. degree in Electrical Engineering from the Democritus University in Greece. He joined the Honeywell Laboratories immediately upon his graduation in January 1997. His expertise is in the areas of Computer Vision beyond the Visible Spectrum, Computational Physiology, and Software Engineering. Dr. Pavlidis published extensively in these areas in major journals and refereed conference proceedings over the past several years. His publication record includes articles in the IEEE Proceedings, New England Journal of Medicine, The Lancet, and Nature. He is also the author of the book "Programming of Cameras and Pan-Tilts with DirectX and Java" by Morgan Kaufmann.
Dr. Pavlidis is also the author of multiple patents, some of which protect multi-million dollar products of Fortune 30 companies (e.g., DVM of Honeywell). Dr. Pavlidis' research work was cited extensively in the scientific literature and has been mentioned frequently by the international media including CNN, The Reuters, The Time magazine, The L.A. Times, and the Discover magazine. Dr. Pavlidis is a Fulbright Fellow, a Senior Member of IEEE, and a member of ACM. Dr. Pavlidis joined the Computer Science Department of the University of Houston in September 2002, where he currently holds the position of Associate Professor.
CPSC 681 Faculty Contact: Ricardo Gutierrez-Osuna (rgutier [at] cs.tamu.edu)
Dr. Ricardo Gutierrez-Osuna, Assistant Professor of Computer Science, Texas A&M University
4:10 p.m., Monday October 10, 2005
Room 124, Bright Building
The integration of gas sensor arrays and pattern-analysis techniques has received much attention in recent years as a low-cost alternative to odor measurement, conventionally carried out with analytical instruments or human panels. When exposed to volatile compounds, the multivariate response of an array of cross-selective sensors can be used as a digital fingerprint, and processed with pattern recognition tools to predict the identity and/or concentration of the volatiles.
To date, most of the pattern recognition for chemical sensor arrays has been based on statistical techniques. The objective of our work is to develop new algorithms inspired by the biological olfactory system. We will present a review of the main computational principles in the olfactory pathway, and describe their application to the processing of sensor-array signals. We will also discuss the correlation of instrument data with sensory analysis from human panels.
Ricardo Gutierrez-Osuna received the B.S. degree in Electrical Engineering from the Polytechnic University of Madrid (Spain) in 1992, and the M.S. and Ph.D. degrees in Computer Engineering from North Carolina State University in 1995 and 1998, respectively. From 1998 to 2002 he served on the faculty at Wright State University. He is currently an assistant professor in the Department of Computer Science at Texas A&M University. Dr. Gutierrez-Osuna is a recipient of the National Science Foundation CAREER Award for his research on machine olfaction with chemical sensors arrays. His research interests include pattern recognition, biological cybernetics, sensor instrumentation, speech-driven facial animation, and mobile robotics.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Rida Bazzi, Associate Professor, Arizona State University
4:10 p.m., Monday October 17, 2005
Room 124, Bright Building
We study ways to restrict or prevent the damage that can be caused in a peer-to-peer network by corrupt entities creating multiple pseudonyms (Sybil attack). We show that it is possible to remotely issue certificates that can be used to test the distinctness of identities. To our knowledge, this is the first work that shows that remote anonymous certification of identity is possible under adversarial conditions. Our certification protocols are based on geometric techniques that establish location information in a fault-tolerant and distributed fashion. They do not rely on a centralized certifying authority or infrastructure that has direct knowledge of entities in the system, and work in Euclidean or spherical geometry of arbitrary dimension. Our protocols tolerate corrupt entities, including corrupt certifiers as well as collusion by certification applicants and certifiers. We consider both broadcast and point-to-point message passing models.
Rida Bazzi is an associate professor in the Computer Science and Engineering department at Arizona State University. He received his PhD and M.Sc. degrees in Computer Science from the Georgia Institute of Technology in 1994. His research is in the general area of reliability in distributed systems including fault tolerance and security. Professor Bazzi was a recipient of an NSF CAREER award.
CPSC 681 Faculty Contact: Jennifer Welch (welch [at] cs.tamu.edu)
Dr. Bradley Jensen, Microsoft
4:10 p.m., Monday October 24, 2005
Room 124, Bright Building
This talk is oriented towards those interested in a high level overview of the Microsoft .NET Compact Framework and developing for mobile devices. An introduction to the basic concepts of the .NET Compact Framework will be presented and it will be shown how the standard .NET Compact Framework integrates with mobile devices and supports mobile application development. Attendees will learn the basics of working with compact user interfaces, how to store and retrieve data on the compact device, and how to deploy mobile applications. This talk will be of interest to those working in embedded systems.
Bradley K. Jensen received his Ph.D. in Business Computer Information Systems from the University of North Texas (UNT), with majors in Business Computer Information Systems and Computer Science. He is a Microsoft Corporation Academic Relationship Manager responsible for Texas, Oklahoma, Arkansas, and Louisiana. Prior to Microsoft, he was an Assistant Professor in Information Technology and Decision Sciences and Assistant Director of the Information Systems Research Center at UNT, and was also President of JMC Consulting Services, an executive management consulting firm which provides strategic and tactical IT consulting services. His research interests include privacy and security, networking, human factors, e-commerce, and document management. Dr. Jensen has been an executive and consultant with more than 20 years of sales, marketing, and IT experience with several Fortune 100 companies.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Santiago Marco, Associate Professor, University of Barcelona
4:10 p.m., Wednesday November 2, 2005
Room 124, Bright Building
Low cost chemical solid-state sensors or miniature spectrometers open new ways for sensing in the field. In the talk, chemical sensing signal processing using bilinear and trilinear models will be introduced. IMS (Ion Mobility Spectrometers) is an established technology for the prevention of terrorist acts or law enforcement, recently miniature ion mobility spectrometers have been presented in the literature. At the talk the scaling behaviour of miniature IMS will be discussed and the options for spectra signal processing will be introduced. In particular some space will be devoted to the options offered by MCR (Multi Curve Resolution) for blind source separation in comparison with SIMPLISMA or PCA. The conditions imposed will be discussed and compared to those used in ICA. Results will show how MCR is able to recover concentration profiles in a variety of settings of noise levels and peak overlapping. Finally, It is well known that drift is always an issue in low cost chemical sensing. In the talk a new procedure for drift counteraction based on multi-way methods will be introduced and compared with component correction (CC).
Dr. Santiago Marco is associate Professor (Profesor Titular) at the Departament d'Electronica of Universitat de Barcelona since 1995. He received the degree in Physics from the Universitat de Barcelona in 1988. From 1989 to 1990 he was working in the electro-optical characterization of deep levels in GaAs. From 1990 to 1993 he was regular visitor of the Centro Nacional de Microelectronica, Bellaterra, Spain. In 1993, he received his Ph.D. (honor award) degree from the Departament de Fisica Aplicada i Electronica, Universitat de Barcelona, for the development of a novel silicon sensor for in-vivo measurements of the blood pressure. In 1994, he was a post-doc researcher at the Department of Electronic Engineering, Universita di Roma 'Tor Vergata', working in Data Processing for Artificial Olfaction. He has published about 50 papers in scientific journals and books, as well as more than 100 conference papers. His current research interests are twofold: chemical instrumentation based on intelligent signal processing and microsystem modeling.
CPSC 681 Faculty Contact: Ricardo Gutierrez-Osuna (rgutier [at] cs.tamu.edu)
Dr. Injong Rhee, Associate Professor, North Carolina State University
4:10 p.m., Wednesday November 9, 2005
Room 124, Bright Building
In this talk, we present the design, implementation and performance evaluation of a hybrid MAC protocol, called Z-MAC, for wireless sensor networks that combines the strengths of TDMA and CSMA while offsetting their weaknesses. Like CSMA, Z-MAC achieves high channel utilization and low-latency under low contention and like TDMA, achieves high channel utilization under high contention and reduces collision among two-hop neighbors at a low cost. A distinctive feature of Z-MAC is that its performance is robust to synchronization errors, slot assignment failures and time-varying channel conditions; in the worst case, its performance always falls back to that of CSMA. Z-MAC is implemented in TinyOS.
This work is a result of collaboration with the following graduate students: Ajit Warrior, Mahesh Aia and JK Min.
Injong Rhee received his PhD in Computer Science from the University of North Carolina at Chapel Hill in 1994. After several years of postdoctoral experience in Warwick University, UK and Emory University, USA, we joined the Computer Science Department of North Carolina State University in 1997 where he is currently associate professor. From 2000 to 2002, Injong was on leave from his university position to found Togabi Technologies, INC, a San Diego based startup that specializes in developing and marketing wireless multimedia applications and services for wireless Internet service providers. He served as CTO/CEO of the company before returning to his academic position in 2003. His main research area is in computer networks and he is interested in developing scalable and practical network protocols for congestion control, multimedia networking and wireless networks.
He is a recipient of NSF CAREER award, 1998.
CPSC 681 Faculty Contact: Jennifer Welch (welch [at] cs.tamu.edu)
Dr. Timothy Darr, Technical Team Lead, 21st Century Technologies, Inc.
4:10 p.m., Wednesday November 16, 2005
Room 124, Bright Building
We present an approach for identifying and responding to asymmetric threats against politically, culturally, and socially diverse adversaries. This approach integrates state-of-the-art game theory and simulation research with proven, powerful graph-matching and Social-Network Analysis (SNA) methods to model and detect asymmetric threat activities through multi-agent adversarial games. Innovative aspects of this approach include: (1) culturally based game-theoretic models of the asymmetric threat, (2) empirical game-theoretic analysis of the asymmetric threat, (3) approximate solutions for intractable games identifying plausible threat strategies and profiles, (4) graph-based threat patterns and SNA signatures derived from the game-theoretic simulation and analysis, and (5) analysis of politically, socially, and culturally diverse adversaries. This approach provides sophisticated technology to a wide audience of potential users for enhanced national security against asymmetric threats resulting in significant improvements in our response capabilities against politically, socially, and culturally diverse adversaries. Additionally, this approach will have a wide range of important defense and commercial applications including terrorist threat detection, industrial espionage detection, financial fraud detection, and business intelligence.
Tim Darr received his B.S.E. in Computer Engineering, M.S.E. in Computer Science and Engineering, and Ph.D. in Computer Science from the University of Michigan. While at the University of Michigan, he studied at the Artificial Intelligence Lab, and held a post-doctoral appointment in the School of Information. Past research interests include distributed design, concurrent engineering and constraint-satisfaction. Current research interests include knowledge representation and acquisition, graph theory, SNA, and agent-based systems. Dr. Darr's publications have appeared in IEEE Expert, AI EDAM, the AI in Design Conference, and the ASME Design Engineering Technical Conference. His research has also appeared in several invited workshops, and he has co-authored chapters in books on concurrent engineering and expert systems. In addition, he was editor for a special issue devoted to configuration design for the journal AI EDAM.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Perakath Benjamin, Knowledge Based Sysstems, Inc.
4:10 p.m., Monday November 21, 2005
Room 124, Bright Building
The seminar will describe innovative methods and tools for building knowledge-based applications. Topics covered will include
As the Vice President for Research and Development at Knowledge Based Systems, Inc. (KBSI), Dr. Perakath Benjamin manages and directs the R&D activities at KBSI. He has over 18 years of professional experience in the design, development, and deployment of advanced systems and applications. Dr. Benjamin has a Ph.D. in Industrial Engineering from Texas A&M University. Dr. Benjamin has been responsible for the development of innovative methods and tools that are being applied extensively throughout industry and government.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Scott Pike, Assistant Professor, Texas A&M University
4:10 p.m., Wednesday November 30, 2005
Room 124, Bright Building
We present theoretical and practical techniques for isolating partial failures in distributed systems to small, local neighborhoods of impact. Specifically, we present scalable algorithms for minimizing the impact of crash faults in a broad class of static resource allocation problems. Our particular lens of investigation focuses on the generalized dining philosophers problem as a fundamental abstraction for distributed resource allocation. Within this domain of inquiry, we construct fault-tolerant algorithms that restrict the scope of failures precipitated by crash faults. Additionally, we prove impossibility results for our techniques and optimality results for our constructions under different models of mutual exclusion and process synchronization. An overarching theme of this work is the central role of locality in the construction of scalable algorithms that support the survivability and availability of distributed systems.
Scott Pike is an Assistant Professor at Texas A&M University. He received his Ph.D. and M.S. degrees in computer science from the Ohio State University in 2004 and 2000, respectively, preceded by a B.A. in philosophy from Yale University in 1996. His research interests focus on distributed computing and software engineering, and, more concretely, on scalable approaches to building agile, adaptive, and survivable components for distributed systems.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Faith Fich, Professor, University of Toronto
4:10 p.m., Wednesday January 18, 2006
Room 124, Bright Building
The atomic snapshot object is an important and well-studied primitive in distributed computing that allows processes to obtain a consistent view of an entire block of shared memory, even if updates are being performed concurrently. This talk will describe applications of this object, present some implementations of it from registers in both asynchronous and synchronous distributed systems, and discuss some recent lower bounds.
Faith Ellen Fich is a Professor of Computer Science at the University of Toronto. She received her Ph.D. in Computer Science from the University of California, Berkeley, in 1982. From 1983 to 1986, she was an assistant professor in the Computer Science Department at the University of Washington. Then she joined the Department of Computer Science at the University of Toronto. Professor Fich is also a member of the Computing Research Association Committee for Women in Computing Research and is the director of the Canadian Distributed Mentor Project. Her research is in the area of complexity theory. She particularly likes to prove lower bounds on the complexity of distributed computing and data structure problems, with the goal of understanding how parameters of various models affect their computational power.
CPSC 681 Faculty Contact: Jennifer Welch (welch [at] cs.tamu.edu)
Dr. Michael Langston, Professor, University of Tennessee
4:10 p.m., Monday January 23, 2006
Room 124, Bright Building
In this talk I will discuss algorithmic methods based on the theory of fixed-parameter tractability. Efficient techniques for problem reduction and massively parallel algorithms for enumeration and search will be discussed. The importance of maintaining a balanced decomposition of the search space turns out to be critical to achieving scalability. Applications to high-throughput computational biology will be stressed, with the analysis of microarray data serving as a prime example. Using mRNA samples obtained from recombinant inbred Mus musculus strains, we solve immense instances of the clique problem to derive sets of putatively co-regulated genes. Techniques for dealing with noisy data are important concerns. The depth of quantitative genetic analysis we can perform is vastly enhanced by combining these results with knowledge of cis-regulatory elements, ontological classifications, and causal structures that may be imposed with quantitative trait locus mapping. A long-term goal is gene regulatory network discovery.
Mike Langston received the PhD in Computer Science from Texas A&M University in 1981, under the most excellent mentorship of Don Friesen. He currently holds the position of Professor of Computer Science at the University of Tennessee, and regularly consults in the Life Sciences, Chemistry, Computer Science and Mathematics Divisions at Oak Ridge National Laboratory. His research is focused primarily on efficient algorithm design, analysis and high performance implementations, with a special emphasis on applications to computational biology. He has authored over 180 refereed publications. His research has been funded by the National Science Foundation, the Department of Defense, the Department of Energy, the National Institutes of Health, and a variety of other federal agencies. He has received numerous awards, most recently the Distinguished Service Prize from the Association for Computing Machinery Special Interest Group on Algorithms and Computation Theory, and the Chancellors Award for Research Creativity at the University of Tennessee.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Michael Gschwind, IBM Master Inventor and Power Architect, IBM Thomas J. Watson Research Center
4:10 p.m., Wednesday February 1, 2006
Room 124, Bright Building
The Cell Broadband Engine was designed to provide an order of magnitude increase in computing performance over desktop systems shipping in 2005. To accomplish this goal in the presence of the diminishing returns of investing in uni-thread performance, the key was to address next generation architecture challenges. Cell addresses the major pressing issues in modern architecture -- the memory wall, the power wall, and the frequency wall -- by shifting the compute paradigm from a uni-processor design to an efficient chip multiprocessor. The Cell Broadband Engine implements a heterogenous chip multiprocessor to deliver a quantum leap in performance by exploiting application parallelism at all levels, i.e., thread level parallelism, instruction level parallelism and data-level parallelism. .
The main source of computing power in the Cell Broadband Engine are 8 Synergistic Processor Elements (SPEs) based on a novel SIMD architecture. SPEs support instruction-level parallelism with a statically scheduled microarchitecture, and data-level paralel parallelism with a pervasively data parallel SIMD architecture. The Synergistic Processor Architecture is based on a streamlined and optimized architecture following RISC principles to implement efficient high-performance pirmitives and use the compiler to layer other computation often implemented in hardware. By streamlining the architecture and exploiting compiler technology to move complexity from runtime to compile time, core efficiency was dramatically increased in terms of power and area usage. This in turn allowed to place one IBM 64-bit Power Architecture(TM) with VMX and 8 SPE cores on a single chip to provide thread-level parallelism to scalar and data-parallel programs. Based on these choices, we will discuss how compiler technology and streamlined hardware allow to deliver superior performance in Cell.
Dr. Michael Gschwind is an IBM Master Inventor, an IBM Power architect and a logic design lead for a future IBM System. He was one of the initiators and a leading contributor to the Cell Broadband Engine system architecture definition as well as a lead architect of the Synergistic Processor architecture. During the definition of the Synergistic Processor architecture, Michael also developed the first Cell Broadband Engine compiler. Michael joined the IBM TJ Watson Research Center, Yorktown Heights, NY, in 1997. He has held leadership positions in several seminal projects, including the DAISY dynamic compilation project where he was a lead architect for the BOA high-frequency statically scheduled architecture, and was a leading contributor to the development of pioneering dynamic compilation techniques. Michael was also a leading contributor to seminal work on power/performance trade-offs in microprocessor designs which formalized the futility of the frequency-centric uniprocessor design approach used in the industry at the time, an insight that had already guided the design of the Cell Broadband Engine.
Michael's contributions to IBM systems and technology have been recognized with several corporate awards. In addition to his contributions to the design and implementation of IBM systems, he is the author of over 75 papers, covering hardware/software co-design, compiler technology, multimedia processing, and high-performance computer architecture, and has received key patents for his inventions in these areas. In addition to his corporate contributions, Michael has been a faculty member at Technische Universitt Wien, Vienna, Austria, and a visiting faculty member at Princeton University where he has taught classes on advanced computer architecture. Michael received PhD and MS degrees in computer science from Technische Universitt Wien, Vienna, Austria.
CPSC 681 Faculty Contact: Lawrence Rauchwerger (rwerger [at] cs.tamu.edu)
Dr. Clay Williams, Manager, Software Quality and Testing, IBM Thomas J. Watson Research Center
4:10 p.m., Monday February 6, 2006
Room 124, Bright Building
The automatic creation of test cases from models or specifications of a system is a long-standing area of research within the software engineering community. The two major approaches are test derivation from algebraic specifications, and test generation from models, which are typically based on finite state machines. This talk will start by discussing common weaknesses in algebraic and finite state-based approaches, and will next propose a new paradigm for generating test cases. At the heart of this paradigm are three innovations: a precise form of use case modeling, a novel approach for automatically creating testing goals based on model analysis, and a new approach to test case generation based on AI planning. These three areas will be discussed in detail using a consistent example to illustrate technical highlights. The talk will conclude with a discussion of extensions and related research.
Clay Williams is a Research Staff Member and the Manager of the Software Quality and Testing Research group at the IBM Watson Research Center. He has a PhD in computer science from Texas A&M University. His scientific interests include software engineering and software quality, complex systems, and bioinformatics.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Yoonsuck Choe, Assistant Professor of Computer Science, Texas A&M University
4:10 p.m., Wednesday February 15, 2006
Room 124, Bright Building
What is available to developmental programs in autonomous mental development, and what should be learned at the very early stages of mental development? Our observation is that sensory and motor primitives are the most basic components present at the beginning, and what developmental agents need to learn from these resources is what their internal sensory states stand for. In this talk, we investigate the question in the context of a simple biologically motivated visuomotor agent. We observe and acknowledge, as many other researchers do, that action plays a key role in providing content to the sensory state. We propose a simple, yet powerful learning criterion, that of invariance, where invariance simply means that the internal state does not change over time. We show that after reinforcement learning based on the invariance criterion, the property of action sequence based on an internal sensory state accurately reflects the property of the stimulus that triggered that internal state. That way, the meaning of the internal sensory state can be firmly grounded on the property of that particular action sequence. The idea was tested on natural images with varying degrees of noise, and the results confirmed our main thesis. We expect the framing of the problem and the proposed solution presented in this work to help shed new light on autonomous understanding in developmental agents.
This is joint work with S. Kumar Bhamidipati and Stuart B. Heinrich.
Note: This talk presents significantly new results compared to the 681 talk I gave in Fall 2003.
Yoonsuck Choe is an assistant professor in the Department of Computer Science at Texas A&M University. He is also serving as the director of the Brain Networks Laboratory in the department. He received his B.S. degree in Computer Science from Yonsei University (Korea) in 1993, and his M.S. and Ph.D. degrees in Computer Sciences from the University of Texas at Austin in 1995 and 2001. His current research areas include computational neuroanatomy, computational modeling of the brain, computational and biological vision, thalamocortical basis of analogy and cortical integration, and autonomous semantics in sensorimotor agents.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Du Li , Assistant Professor of Computer Science, Texas A&M University
4:10 p.m., Monday February 20, 2006
Room 124, Bright Building
The whole human society is a collaborative system. People interact, work, play, fight, and share information with each other as a necessity of life. Today human beings are increasingly more expensive resources than machines. Groupware in general aims to make a "better" utilization of human resources such that some collaboration activities are more productive and successful. The design and evaluation of groupware must consider both social and technical factors as well as their subtle interaction. This talk tells a story of several groupware projects that the speaker and his students experienced. Some of the lessons should be of common interest to faculty and students in related and other areas, simply because any professional stakeholder or activity is contextualized in some collaborative systems in the small and in the large.
Dr. Du Li joined the Department of Computer Science, Texas A&M University as an Assistant Professor in 2000. His research focuses on collaborative systems, addressing a wide spectrum of systems, algorithms, and user interfaces issues. He received a NSF CAREER award in 2002 and has been well established in his research field, computer-supported cooperative work (CSCW). For more information, visit http://www.csdl.tamu.edu/~lidu.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Dezhen Song , Assistant Professor of Computer Science, Texas A&M University
4:10 p.m., Monday March 20, 2006
Room 124, Bright Building
Our broad goal is to combine robotics, cameras, sensors, actuators, and human input to observe and record detailed animal behavior in remote settings.
I'll describe a new project that is just getting started: we are assisting the Cornell team that is searching the Arkansas Big Woods for the elusive Ivory Billed Woodpecker, thought extinct since 1940s but recently sighted. Our goal is defined by the title above.
I'll present recent results on robots collaboratively controlled by humans via networks and describe the issues most relevant to computer vision. The NSF project will investigate the algorithmic foundations for such observatories: new metrics, models, data structures, and algorithms, that will comprise a robust, mathematical framework for collaborative observation. Newly available robotic cameras offer pan, tilt, and extreme zoom capabilities with built-in network servers at low cost. These cameras motivate the Single Frame Selection (SFS) problem, where n users share control of a single robotic camera. I'll present several algorithms, O(n^2 m) for a set of m zoom levels, and O((n + 1/epsilon^3) log^2 n) for an infinite set of zoom levels. The algorithms can be distributed to run in O(nm) time at each client and in O(nlogn) time at the server. The new project will produce working prototypes that will be accessible via the internet to scientists, students, and the public worldwide. Our first online prototype is available for public use at: http://www.c-o-n-e.org/
Ken Goldberg, Judith Donath, Sariel Har-Peled, Vladlen Koltun, and Frank van der Stappen have contributed to this work.
Dezhen Song is an Assistant Professor with Texas A&M University, College Station, Texas, TX, USA. Song received his Ph.D. in 2004 from University of California, Berkeley. Song's primary research area is networked robotics, computer vision, optimization, and stochastic modelling. He received the Kayamori Best Paper Award of the 2005 IEEE International Conference on Robotics and Automation (With J. Yi and S. Ding). http://faculty.cs.tamu.edu/dzsong
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Xiliang Liu, City University of New York
4:10 p.m., Monday April 3, 2006
Room 124, Bright Building
In recent years, bandwidth estimation based on end-to-end packet-train probing has attracted significant research interest. A number of techniques have been proposed to measure the bottleneck bandwidth and cross-traffic rate of an Internet path; however, most of these techniques are based on a fluid queuing model that assumes deterministic cross-traffic arrival at each router and a single-hop network path. In this talk, we present a novel approach to bandwidth estimation that models both bursty (i.e., stochastic) traffic arrival and cases of arbitrarily routed flows in general multi-hop networks. We show that the traditional fluid model usually does not serve as a valid first-order approximation to the stochastic response curve of an end-to-end path and that this difference can cause a significant measurement bias to some of the existing techniques. We also show that a certain portion of the deviation between stochastic and fluid models (and the resulting measurement bias) vanishes as packet-train length increases. We conclude the talk by introducing a new methodology for measuring the bottleneck link capacity and overcoming some of the limitations of previous fluid-based approaches.
Dr. Xiliang Liu received his Ph.D. in computer science from The City University of New York in 2005. His current research interests include network measurement and monitoring, bandwidth estimation, overlay networks, and stochastic modeling.
CPSC 681 Faculty Contact: Dmitri Loguinov (dmitri [at] cs.tamu.edu)
Nancy Amato, Professor of Computer Science, Texas A&M University
4:10 p.m., Monday April 10, 2006
Room 124, Bright Building
In this talk, we describe the PRM framework and give an overview of several PRM variants developed in our group. We describe in more detail our work related to virtual prototyping, computer animation, and protein folding. For virtual prototyping, we show that in some cases a hybrid system incorporating both an automatic planner and haptic user input leads to superior results. For computation animation, we describe new PRM-based techniques for planning for deformable objects and for planning sophisticated group behaviors (flocking and herding). Finally, we describe our application of PRMs to protein folding where we construct a map of the protein's potential landscape which can be used to generate transitional motions of a protein to the native state from unstructured conformations or between specified conformations. More information regarding our work, including movies, can be found at http://parasol.tamu.edu/groups/amatogroup/
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Arch D. Robison, Intel Corporation
4:10 p.m., Wednesday April 12, 2006
Room 124, Bright Building
With multi-core microprocessors becoming ubiquitous, parallel programming for shared-memory machines must evolve from an esoteric specialty to mainstream practice. Though threads are a popular means of shared-memory parallel programming, they are a low-level unstructured construct whose undisciplined use can cause both correctness and performance problems. This talk explains the performance problems, and a solution based on generic programming with a C++ template library. The programmer specifies patterns of logical tasks, not threads, and lets the library schedule the tasks onto processors. The library can deliver a higher level of abstraction and performance than typical hand-threaded code.
Arch D. Robison was the lead developer for KAI C++. Before that, he worked at Shell on massively parallel codes for seismic imaging. He received his Ph.D. from the University of Illinois. He is currently working on a template-based library for multi-core programming.
CPSC 681 Faculty Contact: Lawrence Rauchwerger (rwerger [at] cs.tamu.edu)
Dr. Eun Jung Kim, Assistant Professor of Computer Science, Texas A&M University
4:10 p.m., Wednesday April 19, 2006
Room 124, Bright Building
Widespread use of cluster systems in a diverse set of applications has spurred significant interest in designing such servers, considering performance, scalability, and Quality-of-Service (QoS). However, numerous security loopholes of cluster servers come to the forefront and subsequently the design of secure clusters have recently surfaced as a critical issue.
In this talk, first I will introduce these vulnerabilities from three classical security aspects: availability, confidentiality, and authentication in the InfiniBand Architecture (IBA) framework that may emerge as the de facto standard for designing future system area networks (SANs) for clusters. For better availability of IBA, we propose an efficient implementation method using trap messages. For confidentiality, we encrypt only secret keys to minimize performance degradation. I also present a new authentication mechanism that treats the Invariant CRC (ICRC) field as an Authentication Tag. Finally I will present a deterministic distance packet marking (DDPM) scheme to identify the source nodes generating spoofed IP packets in cluster interconnects. Our simulation results indicate that our methods enhance security in IBA with marginal performance overhead.
Dr. Eun Jung Kim is an assistant professor in the Department of Computer Science at Texas A&M University. Her research interests include Computer Architecture, Power Efficient Systems, Parallel/Distributed Systems, Cluster Computing, and Security. She received a Ph. D degree in the Department of Computer Science and Engineering at the Pennsylvania State University, an MS degree in Computer Science from Pohang University of Science and Technology, Korea and a BS degree in Computer Science from Korea Advanced Institute of Science and Technology, Korea. More information about her research is available at http://faculty.cs.tamu.edu/ejkim.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
Dr. Jaakko Jarvi, Assistant Professor of Computer Science, Texas A&M University
4:10 p.m., Monday May 1, 2006
Room 124, Bright Building
Support for object-oriented programming has become an integral part of mainstream languages, and more recently generic programming has gained widespread acceptance as well. A natural question is how these two paradigms, and their underlying language mechanisms, should interact. One particular design option, that of using subtyping to constrain the type parameters of generic functions, has been chosen in the generics of Java and C#. Certain shortcomings have previously been identified in using subtyping for constraining parametric polymorphism in the context of generic programming. To address these, we propose extending object-oriented interfaces and subtyping to include associated types and constraint propagation. Associated types are type members of interfaces and classes. Constraint propagation allows certain constraints on type parameters to be inferred from other constraints on those parameters and their use in base class type expressions. We present these extensions in the context of C#, describing a translation of the extended features to C#, and providing a formalism proving their safety. The formalism is applicable to other mainstream object-oriented languages supporting F-bounded polymorphism, such as Java.
Jaakko Jarvi is an assistant professor in the Department of Computer Science at Texas A&M University. He has a Ph.D. in Computer Science from the University of Turku, Finland. His research interests include generic programming, programming languages, and software construction in general. He actively participates in the C++ standards committee and is a contributing member of the C++ Boost community. See http://faculty.cs.tamu.edu/jarvi for more information.
CPSC 681 Faculty Contact: Nancy Amato (amato [at] cs.tamu.edu)
