Csaba Szepesvari will be visiting ISIS October 11-13, 2010.  While here, he will doing a presentation called "How to choose cakes (if you must?)--Advice from statistics.

Csaba Szepesvari received his PhD in 1999 from "Jozsef Attila" University, Szeged, Hungary.  He is currently an Associate Professor at the Department of Computing Science of the University of Alberta and a principal investigator of the Alberta Ingenuity Center for Machine Learning.  Previously, he held a senior researcher position at the Computer and Automation Research Institute of the Hungarian Academy of Sciences, where he headed the Machine Learning Group.  Before that, he spent 5 years in the software industry.  In 1998, he became the Research Director of Mindmaker, Ltd., working on natural language processing and speech products, while from 2000, he became the Vice President of Research at the Silicon Valley company, Mindmaker, Inc.  He is the coauthor of a book on non-linear approximate adaptive controllers, a recent short book on Reinforcement Learning, published over 100 journal and conference papers.  He serves as the Associate Editor of IEEE Transactions on Adaptive Control and AI Communications, is on the board of editors of the Journal of Machine Learning Research and the Machine Learning Journal, and is a regular member of the program committee at various machine learning and AI conferences.  His areas of expertise include statistical learning theory, reinforcement learning, and non-linear adaptive control. 

Seminar October 12, 2010, 2:00 pm-4:00 pm, ISIS Large Conference Room

Abstract: How to choose cakes (if you must?)--advice from statistics

Every time you visit a new town, you go to its best confectionary.  Which cake to choose?  Needless to say, cakes are made a little differently in every town.  Should you choose the familiar favorite of yours, or should you try a new one so that you are not missing something very good?  How to choose if there are a very large number of cakes, maybe more than days in your life?  Or even infinite?  Of course, this problem is an instance of the classic multi-armed bandit problem. 

Applications range from project management, pricing products, through calibration of physical processes, monitoring and control of wireless networks, to optimizing website content.  In this talk I will describe some recent results about when the space of options is very large or even infinite with  more or less structure.  I will outline several open problems with varying difficulty.  The talk is based on the joint work with (in chronological order) Peter Auer, Ronald Ortner (Graz, Austria), Yasin Abbasi-Yadkori (U of A), Sarah Filippi, Olivier Cappe, Aurilien Garivier (Telecom ParisTech, CNRS), and Pallavi Arora and Rong Zheng (University of Houston, TX)