/Reinforcement-Learning-for-Game

A project that uses reinforcement learning to train the agent in a game called pyrat, which is developed by Telecom Bretagne.

Primary LanguagePython

Implementation of Deep Q learning algorithm for game application.

An implementation of Deep Q learning with Pytorch with the combination of game theory for Pyrat - a 1vs1 game created by Telecom Bretagne for learning purpose. Win rate: 82% vs defaut machine greedy algorithm.