Decision-Making & Reinforcement Learning. 3 credits

Catalog description

This course is a survey of planning algorithms and the fundamentals of reinforcement learning. It focuses on applications of particular relevance to robot decision-making and game playing. Topics will include heuristic and constraint-based search, Markov Decision Processes, game playing (including Monte Carlo Tree Search, AlphaGo), task planning via PDDL, and core elements of Reinforcement Learning (e.g., TD methods, actor-critic methods) and modern progress and challenges, emphasizing the impact of deep learning and foundation models. Other topics pertaining the complexities of real-world agents (e.g., belief-space planning) may also be discussed. Offered by Computer Science . May not be repeated for credit.