This is the resources and code repository for the Winter of Data Science Project on playing games using Reinforcement Learning.
We will mainly be following Sutton & Barto for the theoretical aspects of Reinforcement Learning and the book Grokking Reinforcement Learning by Miguel Morales for the implementation details.
-
Week 1: Quick introduction to Python and useful Python modules such as NumPy, MatPlotLib, TensorFlow/PyTorch, etc. Implementing a game of Snake in Python which we'll later use to train our model.
-
Weeks 2 and 3: Introduction to Reinforcement Learning from Sutton and Barto, and implementation of few basic algorithms. Using these to train an agent to play the game we made.
-
Week 4: Studying Deep Reinforcement Learning and improving our agaent.
-
Weeks 5 and 6: Implementing Agents to play Atari games from the paper: Playing Atari with Deep Reinforcement Learning by Mnih, Kavukcuoglu et al and from the OpenAI Blog.