Author : Hugo Caselles-Dupré
Disclaimer : several syntax errors can exist in the code. Use it wisely.
Implementation of DeepMind's Deep Q-Network for Reinforcement Learning on OpenAI Gym, using Python with Keras.
In this directory you can find a Jupyter Notebook with the implementation of DQN. The class Experience Replay allows you train the DQN agent on any OpenAI Gym environment.
An example, CartPole-v0, is provided. The neural network is build with Keras on top of Theano (also works with TensorFlow).
If you have any questions : casellesdupre.hugo@gmail.com.