The goal of this class is to equip the graduate (Ph.D. and possibly the M.S. ) students with the basic principles of reinforcement learning (RL). RL is inspired from the way humans and other mammals use to develop strategies/solutions to fulfill their daily tasks. RL is a learning technique that the learner extracts from its experience by interacting with its environment. The following outline will be followed throughout the class:

1. Introduction
2. Evaluative Feedback
3. Reinforcement Learning Problem
4. Dynamic Programming
5. Monte Carlo Methods
6. Temporal-Difference Learning
7. Eligibility Traces (ETs)
8. Generalization and Function Approximation
9. Planning and Learning
10. Hierarchical RL (HRL)
11. RL in Non-Stationary Environments
12. Transfer Learning in RL
13. Multi-Agent RL

• Midterm: 15%
• Projects: 35%
• Final: 50%

• CSE 729 would be helpful.