AlphaZeroGomoku

Pygame and Keras implementation of Gomoku game with simple AlphaZero.

Requirement

• Python 3.6
• Tensorflow-gpu 1.2.0
• Keras 2.1.3
• pygame 1.9.3

Train the model

Run command below to train the model:

python train.py

The alpha/config.py file is used to config the parameters of PolicyValue network, MCTS, game rules and train process.

Run the game

Run command below to run the game:

python gomoku.py

Policy Value Network used

Experiment

Due to the limited computational resources, we train the model with 2000 times self-paly and get a preliminary results.

AI did not show the expected strength on game and I will continue to review the code.

AI vs AI

AI vs Human, human first

Reference

@article{AlphaZero,  
  title=Mastering the Game of Go without Human Knowledge},  
  author={David Silver, Julian Schrittwieser, Karen Simonyan, Ioannis Antonoglou, Aja Huang, Arthur Guez, Thomas Hubert, Lucas Baker, Matthew Lai, Adrian Bolton, Yutian Chen, Timothy Lillicrap, Fan Hui, Laurent Sifre, George van den Driessche, Thore Graepel & Demis Hassabis},
  journal={Nature,550 (7676):354-359},
  year={2017}
}

@article{AlphaZero,  
  title={Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm},  
  author={David Silver, Thomas Hubert, Julian Schrittwieser, Ioannis Antonoglou, Matthew Lai, Arthur Guez, Marc Lanctot, Laurent Sifre, Dharshan Kumaran, Thore Graepel, Timothy Lillicrap, Karen Simonyan, Demis Hassabis},
  journal={arXiv preprint arXiv:1712.01815v1)
  year={2017}
}

Copyright

See LICENSE for details.