UTCS Colloquium/AI: Michael L. Littman Rutgers University Advancing the Theory and Practice of Model-based Reinforcement Learning ACES 6.304 Friday April 20 2007 at 11:00 a.m.

Contact Name: 
Jenna Whitney
Date: 
Apr 20, 2007 11:00am - 12:00pm

There is a signup schedule for this event.

Type of T

alk: UTCS Colloquium/AI

Speaker Name/Affiliation: Michael L. Littm

an/Rutgers University

Date/Time: Friday April 20 2007 11:00 - N

oon

Location (other): ACES 6.304

Host: Peter Stone

T

alk Title: Advancing the Theory and Practice of Model-based Reinforcement

Learning

Talk Abstract:
Reinforcement learners seek to minimize s

ample complexity the amount of experience needed to achieve adequate behav

ior and computational complexity the amount of computation needed per exp

erience. Focusing on these two issues we have been developing theoreticall

y motivated algorithms
that exhibit practical advantages over existing

learning algorithms. I will present some of my lab''s more recent theoretic

al accomplishments as well as some video footage of robots learning.
Speaker Bio:
Michael L. Littman directs the Rutgers Laboratory for Rea

l-Life Reinforcement Learning (RL3) and his research in machine learning ex

amines algorithms for decision making under uncertainty. After earning his

Ph.D. from Brown University in 1996 Littman worked as an assistant profess

or at Duke University a member of technical staff in AT&T''s AI Principles
Research Department and is now an associate professor of computer science
at
Rutgers. Both Duke and Rutgers awarded him teaching awards and his

research has been recognized with three best-paper awards on the topics of

computer crossword solving complexity analysis of planning and efficient

reinforcement learning. He served on the executive council of the American
Association for Artificial Intelligence and is an advisory board member o

f the Journal of Artificial Intelligence Research and an action editor of t

he Journal of Machine Learning Research.