Implementation of Deep Q Learning to Solve Erdos-Selfridge-Spencer Games
Based on the article:
- Install all packages in requirements.txt
- Install the gym environments using
pip install -e gym-defender/
pip install -e gym-attacker/
- You can choose to load multiple types of defender environment : K can be 5, 10, 15 or 20 and potential can be 0.8, 0.9, 0.95, 0.97 or 0.99 for the defender and 1.01, 1.03, 1.05, 1.1 or 1.2 for the attacker. Don't forget to change the name of the environment based on these values and the initial weights for the model.
You will find the main usages of our environnements and agents in a notebook (Notebook.ipynb).
OpenAI Baselines : https://github.com/openai/baselines
Stable Baselines Guide : https://pythonawesome.com/a-fork-of-openai-baselines-implementations-of-reinforcement-learning-algorithms/
- Maxime Bourliatoux
- Maxence Monfort