• Pratik Sahoo
• Prerak Contractor

This is the official repository for the WiDS project of Reinforcement Learning, 2023-24, containing all the relevant resources and assignments.

The focus will be more on algorithmic aspects rather than theory, with the goal being to cover some classical algorithms followed by Deep Reinforcement Learning.

The primary book we will be following is Grokking Deep Reinforcement Learning by Miguel Morales, which can be found in the resources directory.

If you want to dive further into theory behind RL and related topics, you may also refer to Reinforcement Learning: An Introduction by Andrew Barto and Richard S. Sutton.

Spanning across first 2 weeks, we will first be covering the basics of RL problem, terminologies involved, and some classical algorithms to solve it.

• Chapters 2 and 3 of Grokking DRL: Establishes the mathematical foundations and introduces basic algorithms based on Dynamic Programming

• Chapter 4 of Grokking DRL: Introduces Multi-Armed Bandits and related algorithm. You may skim through the later sections if you want.

• Chapter 5 of Grokking DRL: Introduces algorithms for evaluating a policy. Again, you may skim through later sections introducing TD learning.

• Chapter 6 of Grokking DRL: Introduces algorithms to improve agent's performance. This is a very crucial section. Focus especially on Q-Learning.

Will be up soon