From arielpro at cs.cmu.edu Thu Jan 9 00:13:32 2014 From: arielpro at cs.cmu.edu (Ariel Procaccia) Date: Thu, 9 Jan 2014 00:13:32 -0500 Subject: [Intelligence Seminar] Talk by Abe Othman, Jan 20 at 3pm @ GHC 4405 Message-ID: Speaker: Abe Othman Title: Course Match: Large-Scale Equilibrium Approximation for Fair Combinatorial Allocation Time and place: Jan 20, 2014, 3pm, GHC 4405 Abstract: In the combinatorial allocation problem, agents have combinatorial preferences over bundles of possible allocations. The example I will discuss in detail is business school course allocation, where students may have complex preferences over schedules of courses (e.g., two desirable courses could meet at the same time). Other examples of combinatorial allocation include scheduling workers to shifts and scheduling airlines to landing slots. I will begin by giving the intuition why conventional approaches to this problem fail. I will then introduce Course Match, a mechanism that implements a fair allocation while being approximately truthful and efficient. Unfortunately, Course Match requires minimizing a highly discontinuous function in ~400-dimensional space; the overall search problem was recently shown to be PPAD-complete. I will detail the three-stage search technique we used to solve this difficult problem. Crucially, Course Match was designed to be parallelized and scales linearly in the number of processors used. The Wharton school uses Course Match to allocate courses to their MBA students; a recent production run solved seven billion Mixed-Integer Programs to determine schedules. I'll conclude with a discussion of some of the properties of the mechanism that have emerged in practice. Joint work over various papers with Eric Budish (Chicago Booth), Gerard Cachon (Wharton), Judd Kessler (Wharton), Christos Papadimitriou (Berkeley), Aviad Rubinstein (Berkeley), and Tuomas Sandholm (CMU). -------------- next part -------------- An HTML attachment was scrubbed... URL: From dhouston at cs.cmu.edu Mon Jan 20 09:41:21 2014 From: dhouston at cs.cmu.edu (Dana Houston) Date: Mon, 20 Jan 2014 09:41:21 -0500 Subject: [Intelligence Seminar] January 28, 3:30pm: , Presentation by David Sarne Message-ID: <52DD3591.6040301@cs.cmu.edu> INTELLIGENCE SEMINAR JANUARY 28 AT 3:30PM, IN GCH 6115 SPEAKER: DAVID SARNE (Bar-Ilan University) Host: Ariel Procaccia For meetings, contact Pat Loring (sawako at cs.cmu.edu ) UNKNOWING MORE - THE DOWNSIDE OF INFORMATION IN MULTI-AGENT SYSTEMS In many multi-agent systems we find information brokers or information technologies aiming to provide the agents with more information or reduce the cost of acquiring new information. In this talk I will show that better information can hurt: the presence of an information provider, even if the use of her services is optional, can degrade both individual agents' utilities and overall social welfare. The talk will focus on two specific domains: auctions (where the provided information relates to the common value of the auctioned item) and cooperative information gathering (where costly information is shared between the agents). For the first, I'll show that with the information provider in the market, in conflict with classic auction theory, the auctioneer may prefer to limit the number of bidders that participate in the auction and similarly bidders may prefer to have greater competition. Also, bidders' unawareness of the auctioneer's option to purchase the information does not necessarily play into the hands of the auctioneer and, similarly, bidders may sometimes benefit from not knowing that the auctioneer has the option to purchase such information. For cooperative information gathering I'll present three methods for improving overall and individual performance, all based on limiting and constraining information sharing. Along the talk we will also answer questions such as: why bars use dim lights and loud music; ways that charities could benefit from group buying; and why it makes sense to pay someone to over-price information she wants to sell you. BIO David Sarne is a Senior Lecturer in the Department of Computer Science at Bar-Ilan University. He is also the head of the Intelligent Information Agents (IIA) group. He joined Bar-Ilan in Oct. 2007; before this he was a Post-Doc at Harvard University for two years, following several years in the Israeli hi-tech industry. He holds a B.Sc. in Industrial Engineering and an M.Sc. in Information Systems (both from Tel-Aviv University) and a Ph.D. in Computer Science from Bar-Ilan University. -- -------------- next part -------------- An HTML attachment was scrubbed... URL: From arielpro at cs.cmu.edu Mon Jan 20 10:37:35 2014 From: arielpro at cs.cmu.edu (Ariel Procaccia) Date: Mon, 20 Jan 2014 10:37:35 -0500 Subject: [Intelligence Seminar] Reminder: Talk by Abe Othman, today at 3pm @ GHC 4405 Message-ID: Reminder: Abe will give a talk today at 3pm... Ariel On Thu, Jan 9, 2014 at 12:13 AM, Ariel Procaccia wrote: > Speaker: Abe Othman > > Title: Course Match: Large-Scale Equilibrium Approximation for Fair > Combinatorial Allocation > > Time and place: Jan 20, 2014, 3pm, GHC 4405 > > Abstract: > > In the combinatorial allocation problem, agents have combinatorial > preferences over bundles of possible allocations. The example I will > discuss in detail is business school course allocation, where students > may have complex preferences over schedules of courses (e.g., two > desirable courses could meet at the same time). Other examples of > combinatorial allocation include scheduling workers to shifts and > scheduling airlines to landing slots. > > I will begin by giving the intuition why conventional approaches to > this problem fail. I will then introduce Course Match, a mechanism > that implements a fair allocation while being approximately truthful > and efficient. Unfortunately, Course Match requires minimizing a > highly discontinuous function in ~400-dimensional space; the overall > search problem was recently shown to be PPAD-complete. I will detail > the three-stage search technique we used to solve this difficult > problem. Crucially, Course Match was designed to be parallelized and > scales linearly in the number of processors used. > > The Wharton school uses Course Match to allocate courses to their MBA > students; a recent production run solved seven billion Mixed-Integer > Programs to determine schedules. I'll conclude with a discussion of > some of the properties of the mechanism that have emerged in practice. > > Joint work over various papers with Eric Budish (Chicago Booth), > Gerard Cachon (Wharton), Judd Kessler (Wharton), Christos > Papadimitriou (Berkeley), Aviad Rubinstein (Berkeley), and Tuomas > Sandholm (CMU). > -------------- next part -------------- An HTML attachment was scrubbed... URL: From dhouston at cs.cmu.edu Mon Jan 27 10:11:22 2014 From: dhouston at cs.cmu.edu (Dana Houston) Date: Mon, 27 Jan 2014 10:11:22 -0500 Subject: [Intelligence Seminar] January 28, 3:30pm: , Presentation by David Sarne In-Reply-To: <52DD3591.6040301@cs.cmu.edu> References: <52DD3591.6040301@cs.cmu.edu> Message-ID: <52E6771A.6050106@cs.cmu.edu> > INTELLIGENCE SEMINAR > JANUARY 28 AT 3:30PM, IN GCH 6115 > > SPEAKER: DAVID SARNE (Bar-Ilan University) > Host: Ariel Procaccia > For meetings, contact Pat Loring (sawako at cs.cmu.edu > ) > > UNKNOWING MORE - THE DOWNSIDE OF INFORMATION IN MULTI-AGENT SYSTEMS > > In many multi-agent systems we find information brokers or information > technologies aiming to provide the agents with more information or > reduce the cost of acquiring new information. In this talk I will show > that better information can hurt: the presence of an information > provider, even if the use of her services is optional, can degrade > both individual agents' utilities and overall social welfare. The talk > will focus on two specific domains: auctions (where the provided > information relates to the common value of the auctioned item) and > cooperative information gathering (where costly information is shared > between the agents). For the first, I'll show that with the > information provider in the market, in conflict with classic auction > theory, the auctioneer may prefer to limit the number of bidders that > participate in the auction and similarly bidders may prefer to have > greater competition. Also, bidders' unawareness of the auctioneer's > option to purchase the information does not necessarily play into the > hands of the auctioneer and, similarly, bidders may sometimes benefit > from not knowing that the auctioneer has the option to purchase such > information. For cooperative information gathering I'll present three > methods for improving overall and individual performance, all based on > limiting and constraining information sharing. Along the talk we will > also answer questions such as: why bars use dim lights and loud music; > ways that charities could benefit from group buying; and why it makes > sense to pay someone to over-price information she wants to sell you. > > BIO > > David Sarne is a Senior Lecturer in the Department of Computer Science > at Bar-Ilan University. He is also the head of the Intelligent > Information Agents (IIA) group. He joined Bar-Ilan in Oct. 2007; > before this he was a Post-Doc at Harvard University for two years, > following several years in the Israeli hi-tech industry. He holds a > B.Sc. in Industrial Engineering and an M.Sc. in Information Systems > (both from Tel-Aviv University) and a Ph.D. in Computer Science from > Bar-Ilan University. > -- -- -------------- next part -------------- An HTML attachment was scrubbed... URL: From Morganc at lawyers.com Thu Apr 3 08:45:09 2014 From: Morganc at lawyers.com (Morganc at lawyers.com) Date: Thu, 03 Apr 2014 05:45:09 -0700 Subject: [Intelligence Seminar] ..Contact Me. Message-ID: <533CBFF8000057DF@ms12nec.int.opaltelecom.net> (added by postmaster@mail.svcgb1.int.opaltelecom.net) Hello, Good day to you, Kindly read the attached letter and revert back to me. Regards, Barrister Morgan Esq. -------------- next part -------------- A non-text attachment was scrubbed... Name: Proposal.jpg Type: image/jpeg Size: 255101 bytes Desc: not available URL: From mcweng at gmail.com Thu Apr 3 10:56:50 2014 From: mcweng at gmail.com (Weng Ming-cong) Date: Thu, 3 Apr 2014 22:56:50 +0800 Subject: [Intelligence Seminar] intelligence-seminar-announce Digest, Vol 48, Issue 1 In-Reply-To: References: Message-ID: ? 2014?4?3??????? > Send intelligence-seminar-announce mailing list submissions to > intelligence-seminar-announce at mailman.srv.cs.cmu.edu > > To subscribe or unsubscribe via the World Wide Web, visit > > https://mailman.srv.cs.cmu.edu/mailman/listinfo/intelligence-seminar-announce > > or, via email, send a message with subject or body 'help' to > intelligence-seminar-announce-request at mailman.srv.cs.cmu.edu > > You can reach the person managing the list at > intelligence-seminar-announce-owner at mailman.srv.cs.cmu.edu > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of intelligence-seminar-announce digest..." > > > Today's Topics: > > 1. ..Contact Me. (Morganc at lawyers.com ) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Thu, 03 Apr 2014 05:45:09 -0700 > From: Morganc at lawyers.com > To: Recipients > > Subject: [Intelligence Seminar] ..Contact Me. > Message-ID: <533CBFF8000057DF at ms12nec.int.opaltelecom.net > > (added by > postmaster at mail.svcgb1.int.opaltelecom.net ) > Content-Type: text/plain; charset="iso-8859-1" > > Hello, > > Good day to you, Kindly read the attached letter and revert back to me. > > Regards, > Barrister Morgan Esq. > -------------- next part -------------- > A non-text attachment was scrubbed... > Name: Proposal.jpg > Type: image/jpeg > Size: 255101 bytes > Desc: not available > URL: < > http://mailman.srv.cs.cmu.edu/pipermail/intelligence-seminar-announce/attachments/20140403/f9e0e4bb/attachment.jpg > > > > ------------------------------ > > Subject: Digest Footer > > _______________________________________________ > intelligence-seminar-announce mailing list > intelligence-seminar-announce at mailman.srv.cs.cmu.edu > > https://mailman.srv.cs.cmu.edu/mailman/listinfo/intelligence-seminar-announce > > ------------------------------ > > End of intelligence-seminar-announce Digest, Vol 48, Issue 1 > ************************************************************ > -------------- next part -------------- An HTML attachment was scrubbed... URL: From mcweng at gmail.com Thu Apr 3 10:56:28 2014 From: mcweng at gmail.com (Weng Ming-cong) Date: Thu, 3 Apr 2014 22:56:28 +0800 Subject: [Intelligence Seminar] (no subject) Message-ID: -------------- next part -------------- An HTML attachment was scrubbed... URL: From jlentz at cs.cmu.edu Tue May 6 14:24:11 2014 From: jlentz at cs.cmu.edu (Jill Lentz) Date: Tue, 6 May 2014 14:24:11 -0400 Subject: [Intelligence Seminar] May 13, 2014 12:00 noon: Presentation by Michael Bowling Message-ID: <081339BF5B085646911B69780926302C060E2B983955@EXCH-MB-1.srv.cs.cmu.edu> INTELLIGENCE SEMINAR MAY 13 AT 12:00 NOON, IN NSH 1507 (UNUSUAL TIME) SPEAKER: MICHAEL BOWLING (University of Alberta) Host: Ariel Procaccia For meetings, contact Pat Loring (sawako at cs.cmu.edu) ADVERSARIES, ABSTRACTIONS, AND ALGORITHMS The Computer Poker Research Group at the University of Alberta has for over a decade developed the strongest poker playing programs in the world. We have tested them in competition against other programs, winning 20 of 33 events since the inauguration of the AAAI Computer Poker Competition in 2006. We have also tested them against top professional players, becoming the first to beat professional poker players in a meaningful competition in 2008. Our success follows the modus operandi of the very pioneers of game theory: when facing an intractably complex game, abstract the game to a smaller one and reason in that game. "It seems to us...," as Von Neumann and Morgenstern wrote, "its decisive properties will be conserved in our simplified form." Recently, this approach has been shown to be on shaky ground, or rather on no ground at all. In this talk, I will be looking down to see what, if anything, the abstraction methodology can stand on; and what this line of research means for real-world applications where abstraction is step one, as well as applications that do not involve an apparent adversary. BIO Michael Bowling is a Professor of Computing Science at the University of Alberta. His research focuses on artificial intelligence, machine learning, and game theory; and he is particularly fascinated by the problem of how computers can learn to play games through experience. Michael is the leader of the Computer Poker Research Group, which has built some of the strongest poker playing programs in the world. In 2008, one of these programs, Polaris, defeated a team of top professional poker players in two-player, limit Texas Hold'em, becoming the first program to defeat poker pros in a meaningful competition. He also pioneered the Arcade Learning Environment, a testbed for developing artificial intelligence that can exhibit general competence across a variety of domains. His research has been featured on the television programs Scientific American Frontier and National Geographic Today; in print articles in the New York Times and Wired; and twice in exhibits at the Smithsonian Museums in Washington, D.C. Jill M. Lentz Language Technologies Institute School of Computer Science Carnegie Mellon University 6509 Gates Hillman Complex 5000 Forbes Avenue Pittsburgh, PA 15213 jlentz at cs.cmu.edu T: (412)268-1593 F: (412)268-6298 -------------- next part -------------- An HTML attachment was scrubbed... URL: From dhouston at cs.cmu.edu Mon May 12 15:54:28 2014 From: dhouston at cs.cmu.edu (Dana Houston) Date: Mon, 12 May 2014 15:54:28 -0400 Subject: [Intelligence Seminar] May 13, 12:00 noon:, Presentation by Michael Bowling Message-ID: <537126F4.5090505@cs.cmu.edu> INTELLIGENCE SEMINAR MAY 13 AT 12:00 NOON, IN NSH 1507 (UNUSUAL TIME) SPEAKER: MICHAEL BOWLING (University of Alberta) Host: Ariel Procaccia For meetings, contact Pat Loring (sawako at cs.cmu.edu ) ADVERSARIES, ABSTRACTIONS, AND ALGORITHMS The Computer Poker Research Group at the University of Alberta has for over a decade developed the strongest poker playing programs in the world. We have tested them in competition against other programs, winning 20 of 33 events since the inauguration of the AAAI Computer Poker Competition in 2006. We have also tested them against top professional players, becoming the first to beat professional poker players in a meaningful competition in 2008. Our success follows the modus operandi of the very pioneers of game theory: when facing an intractably complex game, abstract the game to a smaller one and reason in that game. "It seems to us...," as Von Neumann and Morgenstern wrote, "its decisive properties will be conserved in our simplified form." Recently, this approach has been shown to be on shaky ground, or rather on no ground at all. In this talk, I will be looking down to see what, if anything, the abstraction methodology can stand on; and what this line of research means for real-world applications where abstraction is step one, as well as applications that do not involve an apparent adversary. BIO Michael Bowling is a Professor of Computing Science at the University of Alberta. His research focuses on artificial intelligence, machine learning, and game theory; and he is particularly fascinated by the problem of how computers can learn to play games through experience. Michael is the leader of the Computer Poker Research Group, which has built some of the strongest poker playing programs in the world. In 2008, one of these programs, Polaris, defeated a team of top professional poker players in two-player, limit Texas Hold'em, becoming the first program to defeat poker pros in a meaningful competition. He also pioneered the Arcade Learning Environment, a testbed for developing artificial intelligence that can exhibit general competence across a variety of domains. His research has been featured on the television programs Scientific American Frontier and National Geographic Today; in print articles in the New York Times and Wired; and twice in exhibits at the Smithsonian Museums in Washington, D.C. -- Dana M. Houston Language Technologies Institute School of Computer Science Carnegie Mellon University 6511 Gates Hillman Complex 5000 Forbes Avenue Pittsburgh, PA 15213 T: (412)268-4717 F: (412)268-6298 -------------- next part -------------- An HTML attachment was scrubbed... URL: From jlentz at cs.cmu.edu Tue Aug 19 14:24:45 2014 From: jlentz at cs.cmu.edu (Jill Lentz) Date: Tue, 19 Aug 2014 14:24:45 -0400 Subject: [Intelligence Seminar] OR Seminar on August 29, 2014 Message-ID: <081339BF5B085646911B69780926302C061F2E3F04A2@EXCH-MB-1.srv.cs.cmu.edu> **Distributed via the Faculty Services mail distribution system.** The following OR seminar has been posted at: http://server1.tepper.cmu.edu/Seminars/seminar.asp?sort=1&short=Y Please take the time to schedule a meeting with the speaker on an individual basis. When you are at the seminar site, proceed to the OR seminar page. To add yourself for a meeting, click on View/Edit Schedule link and then click on the Edit Schedule link. Enter the name portion of your e-mail address (the @andrew.cmu.edu part is not needed) and click Update at bottom of page. ----------------------------------------------------------- Name: Meinoff Sellmann University: Thomas J. Watson Research Center Date: August 29, 2014 Time: 1:30 to 3:00 pm Location: Tepper Faculty Conference Room 322 Title: Automatic Algorithm Configuration Or: How to Win Solver Competitions Without Actually Writing a Solver Abstract: Practically all algorithms have parameters and it is widely recognized that parameter settings often have an essential influence on algorithm performance. In this talk, we give a comprehensive overview of the state- of-the-art in algorithm portfolios and (instance-specific) algorithm tuning. In particular, we present the robust, inherently parallel genetic algorithm GGA for the problem of configuring algorithms automatically. GGA tunes algorithms with categorical, ordinal, and/or continuous parameters based on a training benchmark set of representative input instances. We then consider the problem of instance-specific algorithm selection and review recent advances in algorithm portfolios which are based on non-model-based cost-sensitive multi-classification. Combining the ideas of (instance oblivious) algorithm tuning and algorithm portfolios, we finally arrive at methods for instance-specific algorithm tuning where a high-performance parameterization for a target algorithm is chosen based on the particular input. Extensive numerical experiments evidence the effectiveness of these methods. Most notably, the methods presented in this talk have led to numerous winning entries at five international MaxSAT and SAT solver competitions. Biography: Meinolf Sellmann received his doctorate degree in 2002 from Paderborn University (Germany) and then went on to Cornell University as Postdoctoral Associate. From 2004 to 2010 he held a position as Assistant Professor at Brown University. He now leads a team on AI for Optimization in the cognitive computing department at IBM Watson Research. Meinolf has published over 60 articles in international conferences and journals, served as Conference Chair of CP 2007, PC Chair of CPAIOR 2013, and associate editor of the Informs Journal on Computing. He received an NSF Early Career Award in 2007 and IBM Outstanding Technical Innovation Awards in 2013 and 2014. For four years in a row Meinolf and his team won at international SAT and MaxSAT Solver Competitions, among others two gold medals for the most CPU-time efficient SAT solver for random and crafted SAT instances in 2011, the best multi-engine approach for industrial SAT instances in 2012, the overall most efficient parallel SAT Solver in 2013 (at which point portfolios were permanently banned from the SAT competition), and seven and four first places at the 2013 and 2014 MaxSAT Evaluations. Jill M. Lentz Language Technologies Institute School of Computer Science Carnegie Mellon University 6509 Gates Hillman Complex 5000 Forbes Avenue Pittsburgh, PA 15213 jlentz at cs.cmu.edu T: (412)268-1593 F: (412)268-6298 -------------- next part -------------- An HTML attachment was scrubbed... URL: From jlentz at cs.cmu.edu Mon Aug 25 09:35:45 2014 From: jlentz at cs.cmu.edu (Jill Lentz) Date: Mon, 25 Aug 2014 09:35:45 -0400 Subject: [Intelligence Seminar] Nomination of Speakers for the Intelligence Seminar Message-ID: <081339BF5B085646911B69780926302C06238D3079C3@EXCH-MB-1.srv.cs.cmu.edu> Dear Colleagues: WE ARE LOOKING FOR NOMINATIONS OF SPEAKERS FOR THE INTELLIGENCE SEMINAR. Please let us know the names of people you would like to have as speakers in the Intelligence Seminar series, and whom you would be willing to host. Please email your nominations to iseminar at cs.cmu.edu. Both external and internal nominations, as well as self-nominations, are welcome. A list of already scheduled speakers is posted on the Intelligence Seminar page, www.cs.cmu.edu/~iseminar/. Jill M. Lentz Language Technologies Institute School of Computer Science Carnegie Mellon University 6509 Gates Hillman Complex 5000 Forbes Avenue Pittsburgh, PA 15213 jlentz at cs.cmu.edu T: (412)268-1593 F: (412)268-6298 -------------- next part -------------- An HTML attachment was scrubbed... URL: From jlentz at cs.cmu.edu Mon Sep 22 10:45:35 2014 From: jlentz at cs.cmu.edu (Jill Lentz) Date: Mon, 22 Sep 2014 10:45:35 -0400 Subject: [Intelligence Seminar] [AI Seminar] Sept. 30, 3:30pm: Presentation by Janusz Marecki Message-ID: <081339BF5B085646911B69780926302C06291A6B2A64@EXCH-MB-1.srv.cs.cmu.edu> ARTIFICIAL INTELLIGENCE SEMINAR SEPTEMBER 30 AT 3:30PM, IN GCH 6501 SPEAKER: JANUSZ MARECKI (IBM Research) Host: Ariel Procaccia For meetings, contact Pat Loring (sawako at cs.cmu.edu) PLAYING IN THE DARK: ON SOLVING SINGLE/MULTISTAGE BAYESIAN STACKELBERG GAMES WITH UNKNOWN PLAYER PREFERENCES Recent years have seen a rise in interest in applying game-theoretic methods to real-world domains, such as public surveillance and infrastructure security, wherein one player (the leader) chooses a strategy to commit to and waits for the other player (the follower) to respond. In arriving at optimal leader strategies in these domains, of critical importance is the leader's ability to act, often over prolonged periods of time, despite its limited knowledge of the preferences of the follower. In this talk, I will first present a suite of efficient algorithms for solving single-stage Bayesian Stackelberg Games with distributional uncertainty over follower payoffs. I will then describe an efficient sampling-based algorithm for solving multistage Bayesian Stackelberg Games where follower payoffs can initially be unknown. Finally, I will discuss an extension of the multistage algorithm that equips the leader with the ability to hide its own preferences and deliberately deceive the adversary. BIO Janusz Marecki is a research staff member in the cognitive computing division of IBM T.J. Watson Research. Janusz obtained his Ph.D. in artificial intelligence from the University of Southern California and Dr.Sc. in mathematical modeling from State Scientific and Research Institute of Information Infrastructure in Ukraine. Prior to joining IBM Research, Janusz was a research assistant at the European Laboratory for Nuclear Research, a research associate at the Ukrainian Academy of Sciences, and a lecturer at the Academy of Computer Sciences in Poland. His research interests are in stochastic optimization, decision and game theory, as well as cortical computing for developing high-performance cognitive computing systems of the future. An author of over 90 refereed publications and 7 patents, Janusz is a recipient of a commendation from the Los Angeles Airport Police, a commendation from the Department of Homeland Security, and an Invention Award from the IBM CEO. Jill M. Lentz Language Technologies Institute School of Computer Science Carnegie Mellon University 6509 Gates Hillman Complex 5000 Forbes Avenue Pittsburgh, PA 15213 jlentz at cs.cmu.edu T: (412)268-1593 F: (412)268-6298 -------------- next part -------------- An HTML attachment was scrubbed... URL: From jlentz at cs.cmu.edu Mon Sep 29 10:22:45 2014 From: jlentz at cs.cmu.edu (Jill Lentz) Date: Mon, 29 Sep 2014 10:22:45 -0400 Subject: [Intelligence Seminar] Sept. 30, 3:30pm: Presentation by Janusz Marecki Message-ID: <081339BF5B085646911B69780926302C06291A6B2DB2@EXCH-MB-1.srv.cs.cmu.edu> ARTIFICIAL INTELLIGENCE SEMINAR SEPTEMBER 30 AT 3:30PM, IN GCH 6501 SPEAKER: JANUSZ MARECKI (IBM Research) Host: Ariel Procaccia For meetings, contact Pat Loring (sawako at cs.cmu.edu) PLAYING IN THE DARK: ON SOLVING SINGLE/MULTISTAGE BAYESIAN STACKELBERG GAMES WITH UNKNOWN PLAYER PREFERENCES Recent years have seen a rise in interest in applying game-theoretic methods to real-world domains, such as public surveillance and infrastructure security, wherein one player (the leader) chooses a strategy to commit to and waits for the other player (the follower) to respond. In arriving at optimal leader strategies in these domains, of critical importance is the leader's ability to act, often over prolonged periods of time, despite its limited knowledge of the preferences of the follower. In this talk, I will first present a suite of efficient algorithms for solving single-stage Bayesian Stackelberg Games with distributional uncertainty over follower payoffs. I will then describe an efficient sampling-based algorithm for solving multistage Bayesian Stackelberg Games where follower payoffs can initially be unknown. Finally, I will discuss an extension of the multistage algorithm that equips the leader with the ability to hide its own preferences and deliberately deceive the adversary. BIO Janusz Marecki is a research staff member in the cognitive computing division of IBM T.J. Watson Research. Janusz obtained his Ph.D. in artificial intelligence from the University of Southern California and Dr.Sc. in mathematical modeling from State Scientific and Research Institute of Information Infrastructure in Ukraine. Prior to joining IBM Research, Janusz was a research assistant at the European Laboratory for Nuclear Research, a research associate at the Ukrainian Academy of Sciences, and a lecturer at the Academy of Computer Sciences in Poland. His research interests are in stochastic optimization, decision and game theory, as well as cortical computing for developing high-performance cognitive computing systems of the future. An author of over 90 refereed publications and 7 patents, Janusz is a recipient of a commendation from the Los Angeles Airport Police, a commendation from the Department of Homeland Security, and an Invention Award from the IBM CEO. Jill M. Lentz Language Technologies Institute School of Computer Science Carnegie Mellon University 6509 Gates Hillman Complex 5000 Forbes Avenue Pittsburgh, PA 15213 jlentz at cs.cmu.edu T: (412)268-1593 F: (412)268-6298 -------------- next part -------------- An HTML attachment was scrubbed... URL: