Foundations of Reinforcement Learning

Catalog description

State-of-art intelligent systems often need the ability to make sequential decisions in an unknown, uncertain, possibly hostile environment, by actively interacting with the environment to collect relevant data. Reinforcement Learning is a general framework that can capture the interactive learning setting. This graduate level course focuses on theoretical and algorithmic foundations of Reinforcement Learning. The topics of the course will include: basics of Markov Decision Process (MDP); Sample efficient learning in discrete MDPs; Sample efficient learning in large-scale MDPs; Off-policy policy optimization; Policy gradient methods; Imitation learning & Learning from demonstrations; Contextual Bandits. Throughout the course, we will go over algorithms, prove performance guarantees, and also discuss relevant applications. This is an advanced and theory-heavy course: there is no programming assignment and students are required to work on a theory-focused course project.