Deep Reinforcement Learning

Course assignments of the Deep Reinforcement Learning course offered at New York University during the Fall of 2020.

• Instructor: Prof. Lerrel Pinto
• Course Assistant: Ishan Khanka

The 4 assignments can be found in the 4 folders in this repository root. To quickly summarize the assignments:

1. Solve a standard ML task related to classifying tabular data
2. Solve Open AI's CarRacing-v0 using Immitation Learning. Also, solve the Gambler Problem as an Markov Decision Process [Unavailable].
3. Solve Open AI's CarRacing-v0 using Deep-Q Network. Then solve it using DQN with Priority Experience Replay
4. Solve Open AI's CartPole-v1 using Policy Gradient methods - Reinforce & Proximal Policy Optimization

The 2nd assignment is well documented, so I recommend that one if you have time for only one. Link: https://github.com/SohamTamba/DeepRL/tree/master/2%20-%20Immitation%20Learning/Racing. As the semester progresses, free time become more and more rare :(

Library Versions

• python: 3.7.9
• pytorch: 1.3.1
• torchvision: 0.4.2
• gym: 0.17.3