Reinforcement Learning

CS-5410/ ECO-6410/ MAT-3215/ CS-4411/ ECO-4410 (Monsoon 2026)

Course Description

In autonomous vehicles, robots, and many other physical settings, at any state a number of decisions need to be made. Each decision comes with an immediate reward but more importantly leads to another probabilistically selected state. In reinforcement learning, we learn optimal decisions through trial and error where good decisions are reinforced and bad ones are punished. This learning regime, especially when combined with deep learning, is an incredibly powerful tool for artificial and augmented intelligence. In this course we will study the underlying theory and learn to code-up many practical and intricate applications.

Topics covered:

Markov Chains, Markov Decision Processes, Dynamic Programming, Value and Policy Iteration, Temporal Difference Learning, Q-Learning, functional approximations, Bandit methods, Neural Networks and Deep Q learning, other deep RL methods.

Course Information