IE5571

This course introduces students to modeling and solution methods for problems in sequential decision-making. Applications include games such as Backgammon, Chess, and Go, as well as robotics, traffic control, resource management, and financial trading strategies. We begin with Markov Decision Processes and Dynamic Programming, and then move to model-free Reinforcement Learning (RL) approaches that include Monte Carlo and Temporal Difference learning. We then discuss extensions of these methods to problems of practical interest where it is necessary to employ a supervised learning method to obtain an optimal strategy or policy. Students will obtain hands-on experience by implementing RL methods in a modern programming language such as Python or Julia.

Prereq: Knowledge of multivariable calculus, linear algebra, and probability at the undergraduate level.

000017

Fall Odd Year