Course Websites

Subject offerings of new and developing areas of knowledge in industrial engineering intended to augment the existing curriculum. See Class Schedule or departmental course information for topics and prerequisites. Course Information: Approved for Letter and S/U grading. May be repeated in the same or separate terms if topics vary.

Course description: This course introduces algorithms for reinforcement learning (RL) and learning-based control. The first part of the course (approximately one third of the term) covers foundational topics, including Markov decision processes and their connections to stochastic control, dynamic programming, rollout algorithms and model predictive control, partially observable MDPs and output-feedback control, multi-armed bandits, policy evaluation and policy gradient methods in RL and control. The remainder of the course addresses advanced topics such as function approximations and deep RL, model-based RL and statistical system identification, contextual bandits and contextual RL, online nonstationary RL and control, Monte Carlo Tree Search, imitation learning, reinforcement learning with human feedback (RLHF), safe RL and constrained learning-based control, meta-RL and switching control, and connections to large language models. Coverage of the advanced topics may vary depending on