CITI Distinguished Lecture SeriesThrough its Distinguished Lecture Series, the Computer and Information Technology Institute (CITI) at Rice University strives to bring some of the most notable researchers in computing and information technology to campus. Because computing and information technology cut across virtually all engineering and science disciplines today, the field of potential speakers is large. CITI seeks speakers whose work has been recognized for its impact on society, who have reputations for engaging their audiences, and who will draw an audience from across the Schools of Engineering and Science and the entire Rice research community. Additionally, CITI seeks speakers whose topics will be of interest to members of Houston's larger scientific community with whom Rice researchers interact. Recent speakers in the CITI distinguished lecture series have been:
Ted Selker, MIT Media Lab
Context aware computingEric Grimson, MIT Computer Science
Computer guided surgeryJim Gray, Microsoft Research
On-line science: The World-Wide telescope as a prototype for computational scienceMonty Denneau, IBM Chief Architect for Blue Gene Cyclops Computer
Computing at the speed of life: The BlueGene/Cyclops supercomputerEric Raymond, President, Open Source Initiative
The open source revolution: How software engineering might finally grow upGene Myers, Vice President of Informatics Research at Celera Genomics
Whole genome assemblies of the drosophila and human genomesJohn Sall, Sr. Vice President at SAS Institute
There ought to be a graph for every statisticLarry Matthies, Jet Propulsion Laboratory Machine Vision Group
Trends in computer vision for NASA and DoD mobile robot programsAmbuj Goyal, Vice President, Systems & Software at IBM T.J. Watson Research Center
Transactional Internet: The Coyote Model
Ted Selker
MIT Media Lab
Context aware computing
Thursday, February 5, 2004Abstract
The familiar and useful come from things we recognize. Many of our favorite things’ appearance communicate their use; they show the change in their value though patina. As technologists we are now poised to imagine a world where computing objects communicate with us in-situ; where we are. We use our looks, feelings, and actions to give the computer the experience it needs to work with us.Keyboards and mice will not continue to dominate computer user interfaces. Keyboard input will be replaced in large measure by systems that know what we want and require less explicit communication. Sensors are gaining fidelity and ubiquity to record presence and actions; sensors will notice when we enter a space, sit down, lie down, pump iron, etc. Pervasive infrastructure is recording it.
This talk will cover projects from the Context Aware Computing Group at MIT Media Lab.
About the Speaker
Ted Selker heads the Media Lab's Context-Aware Computing Group. His research has contributed to hundreds of products ranging from notebook computers to operating systems. He is known for the design of the TrackPoint III in-keyboard pointing device now found in Compaq, Fujitsu, HP, IBM, Sony, TI, and other computers; for creating the COACH adaptive agent that improves user performance (Warp Guides in OS/2); and for the design of the 755CV notebook computer that doubles as an LCD projector. While at IBM, Selker built the User Systems Ergonomics Research, or USER, which is known for creating dozens of product visualizations in the form of prototypes and products yearly. Selker and his inventions have received more than 30 awards from publications including PC Magazine, Business Week, and BYTE.Prior to joining MIT faculty in 1999, he worked at IBM's Almaden Research Center, where he became IBM Fellow in 1996. He has served as a consulting professor at Stanford University, taught at Hampshire College, the University of Massachusetts at Amherst, and at Brown University, and worked at Xerox PARC and Atari Research Lab. Selker is the author of 17 patents and 20 papers in refereed journals and conference proceedings.
Eric Grimson
MIT Computer Science
Computer guided surgery
Tueesday, October 7, 2003Abstract
The current trend towards minimally invasive procedures raises an interesting challenge for surgeons—how to execute precise surgeries through small openings with limited view of nearby structures. Recent advances in computer vision are solving this challenge. Knowledge driven segmentation methods provide detailed, patient-specific reconstructions of relevant anatomy. These models allow a surgeon to visualize the site-localizing tumors while highlighting critical structures; and they provide planning tools for optimal approaches to the tumor. Automated registration techniques accurately align the graphical patient reconstruction with actual position in the operating room, so that surgeons can see the positions of their instruments relative to critical nearby structures in real-time, allowing a surgeon to execute minimally invasive surgeries as if the anatomy was completely visible to them. These tools are being used regularly in a range of surgical procedures, and current research aims to extend them to deal with intra-operative deformations.About the Speaker
Eric Grimson is a Professor of Computer Science and Engineering at the Massachusetts Institute of Technology, and holds the Bernard Gordon Chair of Medical Engineering. He also holds a joint appointment as a Lecturer on Radiology at Harvard Medical School and at Brigham and Women's Hospital. Prof. Grimson currently serves as the Education Officer for the Dept. of Electrical Engineering and Computer Science at MIT. He received a B.Sc. (Hons) in Mathematics and Physics from the University of Regina in 1975 and a Ph.D. in Mathematics from MIT in 1980. Prof. Grimson currently heads the Computer Vision Group of MIT's Computer Science and Artificial Intelligence Laboratory, which has pioneered state of the art systems for activity and behavior recognition, object and person recognition, image database indexing, image guided surgery, site modeling and many other areas of computer vision. Recently, his group has been active in applying vision techniques in medicine for image guided surgery and for computational anatomy.
Jim Gray
Microsoft Research
On-line science: The World-Wide telescope as a prototype for computational science
Wednesday, February 5, 2003Abstract
Computational science has historically meant simulation; but, there is an increasing role for analysis and mining of online scientific data. As a case in point, half of the world's astronomy data is public. The astronomy community is putting all that data on the Internet so that the Internet becomes the world's best telescope: it has the whole sky, in many spectra, and in detail as good as the best 2-year-old telescopes. It is useable by all astronomers everywhere. This is the vision of the virtual observatory--also called the World Wide Telescope (WWT). As one step along that path I have been working with the Sloan Digital Sky Survey (especially Alex Szalay of Johns Hopkins) and CalTech to federate their data in web services on the Internet, and to make it easy to ask questions of the database (see http://skyserver.sdss.org). This talk explains the rationale for the WWT, and describes some the computer science challenges of publishing, federating, and mining scientific data.About the Speaker
Jim Gray is part of Microsoft's research group. His work focuses on databases and transaction processing. Jim is active in the research community, is an ACM, NAE, NAS, and AAAS Fellow, and received the ACM Turing Award for his work on transaction processing. He edits a book series on data management, and has been active in building online databases like http://terraService.Net/ and http://skyserver.sdss.org.
Monty Denneau
IBM Chief Architect for Blue Gene Cyclops Computer
Computing at the speed of life: The BlueGene/Cyclops supercomputer
Wednesday, September 25, 2002Abstract Capable of one quadrillion double precision floating-point operations per second, the Blue Gene supercomputer will be used to simulate from first principles the folding of a protein. This calculation requires on the order of 30 sextillion operations and will run for about one year. The machine is being built in the spirit of aggressive simplicity: the instruction set has almost no instructions, each of the 8,000,000 hardware thread units has almost no hardware, and there is only a single part, replicated 32,000 times. We cheerfully accept and use partially good chips. Breaking with traditional packaging, Blue Gene is just 18 inches high, and you can walk on it. This talk will cover the architecture and design of Blue Gene, along with some discussion of appropriate applications.
About the Speaker
Monty Denneau is the system architect for the Blue Gene computer.About the BlueGene/Cyclops Supercomputer* It will be 1,000 times more powerful than DeepBlue, the IBM computer that defeated chess champion Gary Kasparov in 1997 If a PC were 1 inch high, BlueGene would be over 20 miles high–more than 4 times the height of Mount Everest BlueGene’s pipe can download the entire contents of the World Wide Web in less than a second BlueGene’s switching network is equivalent to giving every person in the world two ISDN modems * From UIUC Computer Science Alumni News, Summer 2001.
Eric Raymond
President, Open Source Initiative
The open source revolution: How software engineering might finally grow up
Thursday, March 14, 2002Abstract
Eric Raymond will speak on the history, present-day status and future of open-source software.About the Speaker
Eric S. Raymond is an observer-participant anthropologist in the Internet hacker culture. His research has helped explain the decentralized open-source model of software development that has proven so effective in the evolution of the Internet. His own software projects include one of the Internet's most widely-used email transport programs. Mr. Raymond is also a science fiction fan, a musician, an activist for the First and Second Amendments, and a martial artist with a Black Belt in Tae Kwon Do. His home page is at http://www.tuxedo.org/~esr.
Gene Myers
Vice President of Informatics Research
Celera Genomics
Whole genome assemblies of the drosophila and human genomes
Wednesday, February 21, 2001Abstract
Shotgun sequence assembly is a classic inverse problem: given a set of segments randomly sampled from a target sequence, the problem is to reconstruct the target. Early programs for this problem assisted a user by finding potential overlapping segments which were then assembled by hand. As the programs became progressively more sophisticated the problem was completely solved by the software but still followed by a manual curatorial pass by the users. Until 1995 it was believed that the practical limit on the size of problems that could be solved was on the order of 30 to 50Kbp, due to the intrinsic difficulties posed by repetitive sequence in the target. In 1995 the assembly of a whole genome shotgun dataset for H. Influenza dispelled the notion of such a barrier. Although the process involved significant human curation and bacterial genomes are less repetitive than those of higher organisms, it still portended an economy of effort unmatched by the more laborious map-based approaches then being pursued for large genomes. In 1996, Weber and Myers proposed a whole genome shotgun approach for the human genome suggesting a protocol that involved sampling several individuals in order to simultaneously obtain polymorphism information. Critics claimed that the computation would involve an impossible amount of computer time, that the size and repetitiveness of the genome would confound all attempts at assembly should sufficient computer efficiency be achieved, and that even if an assembly was produced it would be of an extremely poor quality and partial nature.In 1999 the informatics research team at Celera produced an assembly of the Drosophila genome from a whole genome shotgun data set consisting of 3.2 million reads, 72% of which were paired-end reads from 2Kbp and 10Kbp inserts in a 1 to 1.32 mix. The assembly consisted of completely ordered and oriented contigs covering an estimated 97.2% of the genome with only 1630 gaps of average size 1,415bp. The smaller gaps were PCR closed by the Berkeley Drosophila project in a three month period following the publication of the assembly, and the remaining gaps closed in the ensuing 6 months. The assembly is consistent with STS maps and physical clone maps and was compared against 24% of the genome independently sequenced by other groups. The sequence level comparison revealed that the sequence is better than 99.998% accurate within non-repetitive regions of the assembly and 99.62% accurate within repetitive constructs. The basic conclusion is that whole genome assembly is not only feasible but produces a high-quality result that requires comparatively little finishing work.
In this talk, we will cover the approaches to sequencing whole genomes, illustrate the key computational steps of Celera’s whole genome assembler in an attempt to explain what the critics didn’t understand, and describe our current strategies and progress towards a penultimate assembly of the human genome.
About the Speaker
Gene Myers is Vice President of Informatics Research at Celera Genomics where he is building next generation genomic analysis software and an assembler for the whole-genome shotgun sequencing of the Drosophila and Human genomes. He is also currently on leave from the Department of Computer Science at the University of Arizona where he has served on the faculty since receiving his Ph.D. from the University of Colorado in 1981. His research interests include the design of algorithms, pattern matching, computer graphics, and computational molecular biology. His most recent work has focused on algorithms for the central combinatorial problems involved in DNA sequencing, and on a wide range of sequence and pattern comparison problems. Among the tools he has developed are:Dr. Myers is on the editorial boards for the Journal of Computational Biology and BioInformatics, and on the board of the International Society for Computational Biology.
- Blast, a widely used tool for protein similarity searches,
- FAKtory, a system to support DNA sequencing projects,
- Anrep, a pattern matching language for applications in molecular biology, and
- Mac- & PC-Molecule, a molecular visualization tool for Apple and Wintel computers.
This event is sponsored by William J. Rapson, Jr. ‘53
John Sall
Sr. Vice President, SAS Institute
There ought to be a graph for every statistic
Tuesday, April 25, 2000Abstract
For each statistical test, there ought to be a graph showing how the model conspires with the data to arrive at the fit, as well as to reveal patterns and identify points that don’t fit. Sometimes the obvious choice of a graph, such as side-by-side histograms for paired comparisons, is misleading. In the case of means comparisons of unequally-sampled groups, you have to do some clever geometry to show significant differences graphically, using comparison circles. In the case of general linear hypotheses, leverage plots were developed to show the contribution of each effect in the model. Gabriel’s biplots can be extended to 3D to show multivariate relationships. In each case, the goal is to show a graph that has a conceptual foundation relating to the ‘forces’ and ‘energies’ one can visualize from the mechanics of fit. In this way, we promote an understanding of the statistical method, as well as present the results, and provide a stage for making discoveries.About the Speaker
John Sall is cofounder and senior vice president of the SAS Institute, a large software company specializing in statistical analysis. For many years, he was a leader in the development of the SAS System, and more recently leads the development of JMP software, which is used primarily by engineers and research scientists.
Larry Matthies
Jet Propulsion Laboratory Machine Vision Group
Trends in computer vision for NASA and DoD mobile robot programs
Wednesday, December 1, 1999Abstract
NASA and DoD are both funding substantial levels of mobile robot research at present. A key hold-up in both domains has been the very limited ability of robots to sense their environments; however, progress in this area is accelerating considerably. I will survey this progress in three programs that we participate in at JPL: planetary rover research, funded by NASA, portable mobile robots for urban reconnaissance, funded by DARPA, and cross-country mobile robots for reconnaissance in mechanized infantry battalions, funded by the U.S. Army. Autonomous obstacle detection and position estimation are core problems in all of these programs. I will discuss the state of the art and open issues for these problems in each of the above programs, including a discussion of roles played by real-time stereo vision algorithms, compact scanning laser range finders, multispectral cameras, night vision, and visual feature tracking. NASA’s goal is to return samples from Mars by 2008; autonomous rovers are central to achieving this goal and the above technologies are key enablers of the level of autonomy required. For earth-based applications in DoD and elsewhere, these technologies could significantly extend the capabilities of fieldable mobile robots, which today are largely teleoperated, within less than a decade for some applications. Wednesday, December 1, 1999About the Speaker
Dr. Larry Matthies obtained a Ph.D. in Computer Science from Carnegie Mellon University in 1989, then moved to the Jet Propulsion Laboratory, where he is currently supervisor of the Machine Vision Group. He has extensive experience in terrain sensing and obstacle avoidance algorithms for robotic navigation. At JPL, he pioneered the development of real-time algorithms for stereo vision-based obstacle detection for autonomous navigation of robotic vehicles and was a member of the team that developed the Sojourner Mars rover. His group currently has research projects on computer vision for robotic vehicles sponsored by NASA, DARPA, and the U.S. Army; these projects include work on navigation of Mars rovers, asteroid and comet landers, and Earth-based robotic vehicles in urban and cross-country settings. He is a member of the editorial board of the Autonomous Robots journal and an adjunct professor of Computer Science at the University of Southern California.
Ambuj Goyal
Vice President, Systems & Software
IBM T.J. Watson Research Center
Transactional Internet: The Coyote Model
Wednesday, October 13, 1999Abstract
The Internet stretches traditional strict transaction processing concepts in several directions. First, transactions spanning multiple independent organizations may need to address enforcement of pairwise legal agreements rather than a global schema. Second, a new transaction processing paradigm is required that supports different views of units of business for all participants, i.e., service providers as well as end consumers. There may be several related interactions between any two interacting parties dispersed in time creating a long running conversation.We describe the Coyote (COver YOurself Transaction Environment) model for business-to-business (B-B) e-commerce in which
- service contracts between businesses specify the protocol, i.e., the set and sequence of allowable actions;
- an instance of a contract is a long-running conversation, where each action is atomic, the conversation state always moves forward, and recourse actions with possible side-effects are analogous to compensation;
- conversations are coupled by an action in one conversation initiating an action in another conversation.
We show that
- the Coyote model is scalable, since there is no global coordination, and each business is only concerned with contracts with its immediated partners;
- coupled conversations move in synchronism between allowable states specified in the contracts;
- individual and coupled conversations have good termination properties; and
- we show that emerging B-B protocols, such as Open Buying on the Internet (OBI), and RosettaNet, are special cases of this model.
In addition to my technical talk, I will give a brief overview of Computer Science research at IBM.
About the Speaker
Ambuj Goyal heads the division’s worldwide computer science research efforts at IBM Research and is vice president of Systems and Software. In this dual role, he is responsible for setting IBM’s long-term research direction in computer sciences, as well as ensuring the best emerging technologies contribute to IBM’s systems and software products. Dr. Goyal’s main research interests are in the area of fault-tolerant computing, distributed systems and high-performance databases. He has authored more than 50 articles in these fields and has received five outstanding innovation awards from IBM for his work. He was elected an IEEE Fellow for his contributions to the theory and practice of system dependability modeling. Dr. Goyal received his Ph.D. in Electrical Engineering from The University of Texas at Austin, and his Bachelor’s degree from the Indian Institute of Technology at Kanpur. He has been with IBM since 1982.