Personal project to apply deep neural networks to reinforcement learning and create video game AI with Asynchronous Advantage Actor-Critic (A3C) model and Deep Q Networks for OpenAI Gym reinforcement learning environments with TensorFlow/Keras. I only ran it on the CartPole problem but would love to try on more complex domains like Atari with convolutional neural nets when I get a stronger computer.
You must have gym, tensorflow, numpy, and keras. Had some issues running on Windows, especially with gym wrappers but ran well on *nix (I used Ubuntu).
Simply run python [file-name].py!
Asynchronous Methods for Deep Reinforcement Learning, V. Mnih et al., arXiv, 2016.
Playing Atari with Deep Reinforcement Learning, V. Mnih et al., NIPS Workshop, 2013.