From pondabarnes at tti-c.org Wed Oct 3 10:33:26 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Wed Oct 3 10:29:46 2007 Subject: [TTIC Colloquium] Guest Speaker Announcement Message-ID: <002901c805d2$be4f7ad0$e8bf8780@TTIC47> TTI-C Guest Speaker Presented by: Toyota Technological Institute at Chicago Speaker: Ron Dror Speaker's home page: http://people.csail.mit.edu/rondror Date: Tuesday, October 9, 2007 Time: 12:00 Location: TTI-C Conference room, 2nd floor Title: Elucidating Protein Function via Fast, Scalable Molecular Dynamics Simulations Ron Dror, D. E. Shaw Research Abstract: Although molecular dynamics (MD) simulations of biomolecular systems often run for days to months, many events of great scientific interest and pharmaceutical relevance occur on long time scales that remain beyond reach. Such events include functionally important changes in protein structures, folding of proteins to their native three-dimensional structures, and interactions between proteins or between proteins and candidate drug molecules. We present several new algorithms that significantly accelerate parallel MD simulations compared with current state-of-the-art codes. These methods are embodied in a newly developed MD code called Desmond that achieves unprecedented simulation throughput and parallel scalability on commodity clusters. For example, on a standard benchmark, Desmond's performance on a conventional Opteron cluster with 2K processors slightly exceeded the reported performance of IBM's Blue Gene/L machine with 32K processors running its Blue Matter MD code. The ability to perform long MD simulations quickly has allowed us to provide an atomically detailed explanation for the mechanisms of several proteins, including the sodium-proton antiporter NhaA. This is joint work with Kevin Bowers, Edmond Chow, Huafeng Xu, Shy Arkin, Michael Eastwood, Brent Gregersen, Morten Jensen, John Klepeis, Istvan Kolossvary, Mark Moraes, Federico Sacerdoti, John Salmon, Yibing Shan, and David Shaw If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@tti-c.org. For future TTI-C events please visit http://ttic.uchicago.edu/cal/month.php. (TTI-C 1427 E. 60th Street, Chicago, IL). -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20071003/d7ffc3d2/attachment.htm From pondabarnes at tti-c.org Thu Oct 4 10:17:26 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Thu Oct 4 10:13:45 2007 Subject: [TTIC Colloquium] Guest Speaker Announcement Message-ID: <001001c80699$acd85200$e8bf8780@TTIC47> TTI-C Guest Speaker Presented by: Toyota Technological Institute at Chicago Speaker: Sanjeev Khannna Speaker's home page: http://www.cis.upenn.edu/~sanjeev/ Date: Tuesday, October 9, 2007 Time: 11:00 Location: TTI-C Conference room, 2nd floor Title: Disjoint Paths in Networks Abstract: A fundamental problem in combinatorial optimization is the edge-disjoint paths problem (EDP). We are given a network and a collection of source-destination pairs in the network. The goal is to maximize the number of pairs that can be connected by edge-disjoint paths. Even special cases of EDP correspond to non-trivial optimization problems, and the problem becomes NP-hard in very restricted settings. In this talk, we will survey some recent progress on understanding the approximability threshold of EDP and its variants. While the recent developments have essentially resolved the approximability of EDP and related problems in directed graphs, the status of the undirected case remains wide open. We will describe a promising framework for getting much-improved algorithms for undirected EDP when some congestion is allowed. In particular, we will highlight a conjecture whose resolution is strongly tied to the approximability of the undirected case, and describe some results that lend support to this conjecture. If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@tti-c.org. For future TTI-C events please visit http://ttic.uchicago.edu/cal/month.php . (TTI-C 1427 E. 60th Street) -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20071004/e3208a02/attachment.htm From pondabarnes at tti-c.org Tue Oct 9 09:22:41 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Tue Oct 9 09:18:40 2007 Subject: [TTIC Colloquium] RE: Guest Speaker Announcement Message-ID: <003401c80a7f$da5ad790$e8bf8780@TTIC47> REMINDER!! TTI-C Guest Speaker Presented by: Toyota Technological Institute at Chicago Speaker: Ron Dror Speaker's home page: http://people.csail.mit.edu/rondror Date: Tuesday, October 9, 2007 Time: 12:00 Location: TTI-C Conference room, 2nd floor Title: Elucidating Protein Function via Fast, Scalable Molecular Dynamics Simulations Ron Dror, D. E. Shaw Research Abstract: Although molecular dynamics (MD) simulations of biomolecular systems often run for days to months, many events of great scientific interest and pharmaceutical relevance occur on long time scales that remain beyond reach. Such events include functionally important changes in protein structures, folding of proteins to their native three-dimensional structures, and interactions between proteins or between proteins and candidate drug molecules. We present several new algorithms that significantly accelerate parallel MD simulations compared with current state-of-the-art codes. These methods are embodied in a newly developed MD code called Desmond that achieves unprecedented simulation throughput and parallel scalability on commodity clusters. For example, on a standard benchmark, Desmond's performance on a conventional Opteron cluster with 2K processors slightly exceeded the reported performance of IBM's Blue Gene/L machine with 32K processors running its Blue Matter MD code. The ability to perform long MD simulations quickly has allowed us to provide an atomically detailed explanation for the mechanisms of several proteins, including the sodium-proton antiporter NhaA. This is joint work with Kevin Bowers, Edmond Chow, Huafeng Xu, Shy Arkin, Michael Eastwood, Brent Gregersen, Morten Jensen, John Klepeis, Istvan Kolossvary, Mark Moraes, Federico Sacerdoti, John Salmon, Yibing Shan, and David Shaw If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@tti-c.org. For future TTI-C events please visit http://ttic.uchicago.edu/cal/month.php. (TTI-C 1427 E. 60th Street, Chicago, IL). -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20071009/f41e99ff/attachment-0001.htm From pondabarnes at tti-c.org Tue Oct 9 09:23:31 2007 From: pondabarnes at tti-c.org (Ponda Barnes) Date: Tue Oct 9 09:19:30 2007 Subject: [TTIC Colloquium] RE: Guest Speaker Announcement Message-ID: <003901c80a7f$fa47e840$e8bf8780@TTIC47> Reminder!! TTI-C Guest Speaker Presented by: Toyota Technological Institute at Chicago Speaker: Sanjeev Khannna Speaker's home page: http://www.cis.upenn.edu/~sanjeev/ Date: Tuesday, October 9, 2007 Time: 11:00 Location: TTI-C Conference room, 2nd floor Title: Disjoint Paths in Networks Abstract: A fundamental problem in combinatorial optimization is the edge-disjoint paths problem (EDP). We are given a network and a collection of source-destination pairs in the network. The goal is to maximize the number of pairs that can be connected by edge-disjoint paths. Even special cases of EDP correspond to non-trivial optimization problems, and the problem becomes NP-hard in very restricted settings. In this talk, we will survey some recent progress on understanding the approximability threshold of EDP and its variants. While the recent developments have essentially resolved the approximability of EDP and related problems in directed graphs, the status of the undirected case remains wide open. We will describe a promising framework for getting much-improved algorithms for undirected EDP when some congestion is allowed. In particular, we will highlight a conjecture whose resolution is strongly tied to the approximability of the undirected case, and describe some results that lend support to this conjecture. If you have any questions or would like to meet the speaker, please contact Ponda Barnes at pondabarnes@tti-c.org. For future TTI-C events please visit http://ttic.uchicago.edu/cal/month.php . (TTI-C 1427 E. 60th Street) -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20071009/3d4c9f90/attachment.htm From cnovak at tti-c.org Fri Oct 19 15:30:19 2007 From: cnovak at tti-c.org (Christina Novak) Date: Fri Oct 19 15:25:50 2007 Subject: [TTIC Colloquium] Guest Speaker- Avi Pfeffer (TTI-C) Message-ID: <000101c8128e$de684310$a9bf8780@cnovakHBRQFD1> TTI-C Guest Speaker Presented by: Toyota Technological Institute at Chicago Speaker: Avi Pfeffer Speaker's home page: http://www.eecs.harvard.edu/~avi/ Date: Tuesday, October 23, 2007 Time: 2:30pm Location: TTI-C Conference room, 2nd floor Title: Modeling the Reasoning of Agents in Games Abstract: Why do agents (people or computers) do things in strategic situations? Answering this question will impact how we build computer systems to assist, represent or interact with people in interactions with other agents such as negotiations and resource allocation. We identify four reasoning patterns that agents might use: choosing an action for its direct effect on the agent's utility, attempting to manipulate another agent, signalling information to another agent that the first agent knows, or revealing or hiding information from another agent that the first agent itself does not know. We present criteria that characterize each reasoning pattern as a pattern of paths in a multi-agent influence diagram, a graphical representation of games. We define a class of strategies in which agents do not make unmotivated distinctions, and show that if we assume all agents play these kinds of strategies, our categorization of reasoning patterns is complete and captures all situations in which an agent has reason to make a decision. We then study how people use two reasoning patterns in a particular negotiation game. We use machine learning to learn models of people's play, and embed our learned models in computer negotiators. We find that negotiators that use our learned model outperform classical game-theoretic agents and also outperform people. Finally, we learn models of the way people's behavior changes in ongoing interactions with the same agent, particularly the degree to which retrospective (rewarding or punishing past behavior) and prospective (attempting to induce future good behavior) reasoning play a role. If you have any questions or would like to meet the speaker, please contact David McAllester at mcallester at tti-c.org. For future TTI-C events please visit http://ttic.uchicago.edu/cal/month.php. (TTI-C 1427 E. 60th Street, Chicago, IL). -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20071019/fff02fbc/attachment.htm From ghamburg at tti-c.org Mon Oct 29 10:41:34 2007 From: ghamburg at tti-c.org (Gary Hamburg) Date: Mon Oct 29 10:36:24 2007 Subject: [TTIC Colloquium] Talk Message-ID: <005101c81a4a$91a941f0$a2bf8780@ghamburg5566PD1> What Marina Meila: "Consensus ranking under the exponential model When Thu Nov 1 10am - 11am Where TTI-C Conference Room Marina Meila University of Washington mmp@stat.washington.edu This talk is concerned with summarizing -- by means of statistical modeling -- of data that expresses preferences. This data is typically a set of rankings of n items by a panel of experts; the simplest summary is the "consensus ranking", or the "centroid" of the set of rankings. Such problems appear in many tasks, ranging from combining voter preferences to boosting of search engines. We study the problem in its more general form of estimating a parametric model over permutations, known as the Generalized Mallows (GM) model. The talk will present a new exact estimation algorithm, non-polynomial in theory, but extremely effective in comparison with existing algorithms. From a statistical point of view, we show that the GM model is an exponential family, and introduce the conjugate prior for this model class. Then we introduce the infinite GM model, corresponding to "rankings" over an infinite set of items, and show that this model is both elegant and of practical significance. Finally, the talk will touch upon the subject of multimodal distributions and clustering. Joint work with: Bhushan Mandhani, Le Bao, Kapil Phadnis, Arthur Patterson and Jeff Bilmes Host: Nati Srebro -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20071029/82811a0a/attachment-0001.htm From ghamburg at tti-c.org Mon Oct 29 10:42:47 2007 From: ghamburg at tti-c.org (Gary Hamburg) Date: Mon Oct 29 10:37:39 2007 Subject: [TTIC Colloquium] Talk Message-ID: <005601c81a4a$bd0395d0$a2bf8780@ghamburg5566PD1> Subject: Talk Announcement: Peter Dinda on 10/30, 2pm @ TTI-C >> Conference room >> >> Dear All, >> >> Peter Dinda from Northwestern University will be speaking at TTI-C >> this at 2pm this Tuesday. Please send email to umut@tti-c.org if you >> would like to meet him. >> >> >> ** Speaker: Peter Dinda, Northwestern University >> ** Title: The User in Experimental Computer Systems Research >> ** When: 10/30, Tuesday, at 2pm >> ** Where: TTI-C Conference room (1427 E 60th Street, Chicago IL) >> >> ** Abstract: >> Experimental computer systems research typically ignores the end- >> user, modeling him, if at all, in overly simple ways. We argue that >> this (1) results in inadequate performance evaluation of the systems, >> and (2) ignores opportunities. We summarize our experiences with (a) >> directly evaluating user satisfaction and (b) incorporating user >> feedback in different areas of client/server computing, and use our >> experiences to motivate principles for that domain. Specifically, we >> report on user studies to measure user satisfaction with resource >> borrowing and with different clock frequencies in desktop computing, >> the development and evaluation of user interfaces to integrate user >> feedback into scheduling and clock frequency decisions in this >> context, and results in predicting user action and system response in >> a remote display system. We also present initial results on >> extending our work to user control of scheduling and mapping of >> virtual machines in a virtualization-based distributed computing >> environment. We then generalize (a) and (b) as recommendations for >> incorporating the user into experimental computer systems research. >> >> If time permits, I will also briefly describe other new research >> directions and opportunities in my research group. >> >> ** Bio: >> Peter Dinda is an associate professor in the Department of Electrical >> Engineering and Computer Science at Northwestern University, and also >> affiliated with the Northwestern Institute on Complex Systems and the >> Center for Ultrascale Computing and >> Information Security. He holds a B.S. in electrical and computer >> engineering from the University of Wisconsin and a Ph.D. in computer >> science from Carnegie Mellon University. His research has >> two main threads. The first thread is online measurement, >> modeling, and prediction of the dynamic behavior of resources, >> applications, and users. The second thread encompasses >> virtualization technologies, and, more specifically, their use in >> adaptive systems. Detailed information can be found on pdinda.org. >> -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20071029/bef0d246/attachment.htm From ghamburg at tti-c.org Mon Oct 29 10:44:32 2007 From: ghamburg at tti-c.org (Gary Hamburg) Date: Mon Oct 29 10:39:18 2007 Subject: [TTIC Colloquium] Talk Message-ID: <005b01c81a4a$fb8af870$a2bf8780@ghamburg5566PD1> This talk is sponsored by the University of Chicago and TTI-C and will be held in 251 Ryerson at 2:30 PM on Friday, November 9. The contact for this event is Steve Smale (834-2510) smale@tti-c.org. The large-scale structure of real-world networks Mark Newman Department of Physics and Center for the Study of Complex Systems University of Michigan Many systems take the form of networks: the Internet, the World Wide Web, social networks, citation networks, metabolic networks, food webs, and neural networks are just a few examples. In this talk I will show some recent empirical data for these and other networks and discuss how we can discover and understand their large-scale structure and its implications. The problem is that many networks are too large to visualize in their entirety, so to understand what they "look like" we need algorithmic or statistical techniques to pick useful patterns out of large network data sets. I will describe recent work on several methods that attempt to detect structural features such as clustering and hierarchy using spectral and other techniques. I will give a variety of illustrative applications throughout the talk. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://ttic.uchicago.edu/pipermail/colloquium/attachments/20071029/afe8a19d/attachment.htm From nati at uchicago.edu Wed Oct 31 23:53:18 2007 From: nati at uchicago.edu (Nathan Srebro) Date: Wed Oct 31 23:47:56 2007 Subject: [TTIC Colloquium] TODAY (Thursday) 10AM: Marina Meila "Consensus ranking under the exponential model" Message-ID: <154084920710312253r1fc17d6co713b6dba0230dde8@mail.gmail.com> Consensus ranking under the exponential model Marina Meila, University of Washington Thu Nov 1 10am - 11am, TTI-C Conference Room This talk is concerned with summarizing -- by means of statistical modeling -- of data that expresses preferences. This data is typically a set of rankings of n items by a panel of experts; the simplest summary is the "consensus ranking", or the "centroid" of the set of rankings. Such problems appear in many tasks, ranging from combining voter preferences to boosting of search engines. We study the problem in its more general form of estimating a parametric model over permutations, known as the Generalized Mallows (GM) model. The talk will present a new exact estimation algorithm, non-polynomial in theory, but extremely effective in comparison with existing algorithms. From a statistical point of view, we show that the GM model is an exponential family, and introduce the conjugate prior for this model class. Then we introduce the infinite GM model, corresponding to "rankings" over an infinite set of items, and show that this model is both elegant and of practical significance. Finally, the talk will touch upon the subject of multimodal distributions and clustering. Joint work with: Bhushan Mandhani, Le Bao, Kapil Phadnis, Arthur Patterson and Jeff Bilmes Host: Nati Srebro