Description

The significantly expanded and updated second edition of a widely used introduction to reinforcement learning, one of the most active areas of research in artificial intelligence.

Reinforcement Learning by Richard S. Sutton and Andrew G. Barto presents a clear and accessible introduction to the fundamental concepts and algorithms behind reinforcement learning. The field focuses on how an agent learns to make decisions by interacting with an uncertain environment and working to maximize the rewards it receives over time.

This extensively revised second edition introduces new topics, updates existing material, and expands coverage of important developments in the field. As with the first edition, it emphasizes core online learning algorithms, while more mathematical discussions are presented separately in shaded sections for easier reading.

Part I develops the foundations of reinforcement learning within the tabular case, where exact solutions can be obtained. It introduces essential concepts and a range of algorithms, including several additions new to this edition, such as Upper Confidence Bound (UCB) methods.