Dynamic Programming and Reinforcement Learning

Enrollment Requirements: Graduate Prerequisite: ENGEK 500 and (ENGEC 402 or ENGEC 501 or ENGME 501 or ENGSE 501) Introduction to sequential decision making via dynamic programming. The principle of optimality as a unified approach to optimal control of dynamic systems and Markovian decision problems. Applications from control theory and operations research include linear-quadratic problems, the discrete Kalman Filter, inventory control, network, investment, and resource allocation models. Adaptive control and numerical solutions through successive approximation and policy iteration, suboptimal control, and neural network applications involving functional approximations and learning. Same as ENGEC 710 and ENGSE 710. Students may not receive credits for both.