MSAIL Deep Reinforcement Learning Project

This is a brief guide for deep reinforcement learning projects in MSAIL 2020-2021.

Set Up

We'll be using OpenAI's Gym environment for this project.

1. OpenAI Gym - The Gym main website
2. OpenAI Gym Github Repo - The Raw code of Gym in Github
3. OpenAI Gym Documentation - Documentation, with installation instruction

Resources

• Deeplizard RL Module - some combination of theory and OpenAI Gym
• Deep Reinforcement Learning Series (Jonathan Hui) - A deep RL series on Medium that covers a lot of important topics (DQNs, Policy Gradients, etc)
• Deep RL Course - A series with a more implementation heavy approach: they teach Deep RL concepts through implementing agents for different games (Space Invaders, Doom, Sonic)

More Resources: All of these Github repos have links to a ton of useful RL content (other tutorials, courses, textbooks, blog posts, research papers, etc)

• Roboserg Repo
• Dennybritz Repo
• Aikorea Repo

Acknowledgements

Thanks to Qianbo Yin (Grayson) for compiling the first iteration of this resource guide

Michigan Student Artificial Intelligence Lab (MSAIL) 2020-2021