Leslie Kaelbling

Leslie Kaelbling discusses the concept of "Rational Learning" in reinforcement learning, examining how agents can make principled decisions about what to learn and how to learn efficiently. The talk explores the intersection of Bayesian reasoning and RL, addressing how agents can balance exploration and exploitation while maintaining computational tractability.

Leslie Pack Kaelbling is the Panasonic Professor of Computer Science and Engineering at MIT. She has made fundamental contributions to decision-making under uncertainty, including partially observable Markov decision processes (POMDPs), hierarchical planning, and robot learning.

She is a fellow of AAAI and has received the IJCAI Computers and Thought Award. Her current research focuses on making robots that are competent and robust in performing complex tasks in real environments.