Talks & Seminars
Title: Learning Methods for Sequential Decision Making in Practice
Mr. Shivaram Kalyanakrishnan, Dept. of Computer Science, University of Texas at Austin
Date & Time: December 24, 2010 11:30
Venue: Conference Room, 01st Floor, C Block, Kanwal Rekhi Bldg.
From controlling elevators to trading stocks and playing board games, sequential decision making problems occur widely in practice. In this talk, I will briefly describe the basic ideas behind reinforcement learning (RL), a class of methods that is used to perform sequential decision making based on experience. I will proceed to describe the main challenges in scaling existing RL algorithms to complex applications, and describe various ways in which my research has successfully addressed these challenges. On the practical side, I will demonstrate several successes from the domain of robot soccer. On the theoretical side, I will present novel results obtained within the context of multi-armed bandits, a well-studied abstraction of sequential decision making tasks. Finally, I will specify some open problems for future research in the exciting and increasingly important area of reinforcement learning.
Speaker Profile:
Shivaram Kalyanakrishnan is currently a Ph.D. candidate in the Computer Science department at the University of Texas at Austin. After obtaining his B.Tech. from the Indian Institute of Technology Madras in 2004, he entered graduate school to pursue research in the fields of artificial intelligence and machine learning. His primary interests include reinforcement learning, bandit algorithms, agents and multiagent systems, and humanoid robotics. He has extensively used robot soccer as a test domain for his research, and he actively contributes to initiatives such as RoboCup and the Reinforcement Learning competitions. He has published papers in a variety of areas, and has received the Best Student Paper Award at the RoboCup International Symposium on two occasions.
List of Talks


Faculty CSE IT
Forgot Password
    [+] Sitemap     Feedback