トップ   差分 バックアップ リロード   一覧 単語検索 最終更新   ヘルプ   最終更新のRSS



日時: 2011年2月4日(金) 16:00-17:00

会場: 西8号館W棟10階W1008号室

講演者: Martin M�ller 氏 (Department of Computing Science, University of Alberta)

題目: Resource-Constrained Planning: A Monte-Carlo Random Walk Approach


A ubiquitous feature of planning problems is the need to economize limited resources such as fuel or money. Most state of the art planners use search based on relaxation heuristics, which have a fatal flaw for solving critically resource-constrained problems: they basically ignore resource consumption. To address this, one can try to devise better heuristics. Herein we explore the alternative approach changing the nature of the search instead. We devise two improvements to the Monte-Carlo Random Walk method of Nakhost and M�ller: Smart Restarts and On-Path Search Continuation. Systematic experiments with these new methods and with previous planners show how performance depends on problem constrainedness. In critically resource-constrained problems, the new method outperforms previous planners by a large margin. It is also competitive in other, not resource constrained standard benchmarks.



日時: 2010年10月28日(木) 13:20-14:50

会場: 西8号館W棟8階W809号室

講演者: Chia-Li Wang 氏 (Department of Applied Mathematics, National Dong Hwa University)

題目: Simulation of Ruin Probability with Heavy-tail Claims


One of the key issues of today's simulation literature is the estimation of rare events. The difficulty is that the ratio of simulation variance to the square of the probability diverges as the probability goes to zero. To estimate the ruin probability in insurance risk analysis, if the distribution of claims has a light tail, a solution was to use importance sampling by performing an exponential change of measure. However, if it has a heavy tail, exponential moment does not exist so the above method is not applicable. In this talk, we will demonstrate and discuss various attempts for solving this difficulty, which includes conditional approach, order statistics approach and control variate approach.


日時: 2010年11月15日(月) 11:00-12:00

会場: 西8号館W棟8階W809号室

講演者: Michael Bowling 氏 (Department of Computing Science, University of Alberta)

題目: AI After Dark: Computers Playing Poker


The game of poker presents a serious challenge for artificial intelligence. The game is essentially about dealing with many forms of uncertainty: unobservable opponent cards, undetermined future cards, and unknown opponent strategies. Coping with these uncertainties is critical to playing at a high-level. In July 2008, the University of Alberta's poker playing program, Polaris, became the first to defeat top professional players at any variant of poker in a meaningful competition. In this talk, I'll tell the story of this match interleaved with the science that enabled Polaris's accomplishment.


日時: 2010年9月9日(木) 14:00-15:00

会場: 西8号館W棟8階W809号室

講演者: Anders Hanson 氏 (Department of Electrical Engineering, Linkopings Universitet)

題目: A Structure Exploiting Preprocessor for Semidefinite Programs derived from the Kalman-Yakubovich-Popov Lemma


Semidefinite programs derived from the Kalman-Yakubovich-Popov (KYP) lemma are quite common in control and signal processing applications. The programs are often of high dimension which makes them hard or impossible to solve with general-purpose solvers. Here we present a customized preprocessor, KYPD, that utilizes the inherent structure of this particular optimization problem. The key to an efficient implementation is to transform the optimization problem into an equivalent semidefinite program. This equivalent problem has much fewer variables and the matrices in the linear matrix inequality constraints are of low rank. KYPD can use any primal-dual solver for semidefinite programs as an underlying solver.


日時: 2010年7月6日(火) 16:30-17:30

会場: 西8号館W棟8階W809号室

講演者: Tien-Yien Li 氏 (Department of Mathematics, Michigan State University)

題目: The story about chaos


The talk mainly contains my personal contacts with those heavy weights in Chaos, including Japanese Attractor and its author Ueda of course.

講演者の Tien-Yien Li 教授


は、Chaos に関する有名な論文 T.-Y. Li and J. A. Yorke, Period three implies chaos. Amer. Math. Monthly, Vol. 82, pp. 985-992, 1975. の著者です。 Dynamical Systems and Numerical Analysis の分野で多くの優れた 業績を上げています。今回は, Chaos 誕生に関するお話をお願いしました。 7月2日から7月7日まで、東工大に滞在します。Li 教授と懇談したい方は



連絡先: 小島政和 kojima@is.titech.ac.jp


日時: 2010年5月6日(木) 13:20-14:50

会場: 西8号館W棟8階W809号室

講演者: Lang Wu 氏 (University of British Columbia)

題目: ''Joint inference of longitudinal data and survival data in the presence of measurement errors''


We consider a joint model for a nonlinear mixed effects model and a survival model in the presence of covariate measurement errors. We estimate all model parameters simultaneously based on Laplace approximation as well as the commonly-used Monte-Carlo EM algorithm. The models and methods are motivated and illustrated by a study on HIV viral dynamics.