jasmine0122/othello_ltbel

A AlphaZero implementation for Othello 10x10 / Reversi using C++

Othello_LTBeL

Features

• Tree parallelization Monte-Carlo tree search
• Lock-free multi-thread with virtual loss
• MCTS-Minimax hybrids
• Residual neural network evaluation based on Reinforcement learning
• Large-batch inference
• Using TensorFlow C++ API
• Efficient implementation in C++

Award

• International Computer Games Association 2018 Computer Olympiad - Othello10x10 Bronze
• Taiwanese Association for Artificial Intelligence 2018 Computer Game Tournaments - Othello10x10 Gold

Requirements:

• bazel 0.15+
• tensorflow 1.0+
• CUDA
• cuDNN

Installation

1. Install bazel
2. Build tensorflow from source
3. Install Othello_LTBeL

git clone https://github.com/Es1chUbJyan9/Othello_LTBeL.git
mv -r Othello_LTBeL/ tensorflow/

Usage

• Play game

bash Run_Game.sh

• Create training data (about 5000 min)

bash Create_History.sh

References

• Silver, D., Schrittwieser, J., Simonyan, K., Antonoglou, I., Huang, A., Guez, A., ... & Chen, Y. (2017). Mastering the game of Go without human knowledge. Nature, 550(7676), 354.
• Silver, D., Hubert, T., Schrittwieser, J., Antonoglou, I., Lai, M., Guez, A., ... & Lillicrap, T. (2017). Mastering chess and shogi by self-play with a general reinforcement learning algorithm. arXiv preprint arXiv:1712.01815.
• Liskowski, P., Jaskowski, W. M., & Krawiec, K. (2018). Learning to play Othello with deep neural networks. IEEE Transactions on Games.
• Chaslot, G. M. B., Winands, M. H., & van Den Herik, H. J. (2008, September). Parallel monte-carlo tree search. In International Conference on Computers and Games (pp. 60-71). Springer, Berlin, Heidelberg.
• Hingston, P., & Masek, M. (2007, September). Experiments with Monte Carlo Othello. In Evolutionary Computation, 2007. CEC 2007. IEEE Congress on (pp. 4059-4064). IEEE.
• Baier, H., & Winands, M. H. (2015). MCTS-minimax hybrids. IEEE Transactions on Computational Intelligence and AI in Games, 7(2), 167-179.
• Liu, Y. C., & Tsuruoka, Y. (2016). Asymmetric Move Selection Strategies in Monte-Carlo Tree Search: Minimizing the Simple Regret at Max Nodes. arXiv preprint arXiv:1605.02321.
• Rosenbloom, P. S. (1982). A world-championship-level Othello program. Artificial Intelligence, 19(3), 279-320.
• Buro, M. (1997). An evaluation function for othello based on statistics. Technical Report 31, NEC Research Institute.
• Buro, M. (1995, March). Logistello: A strong learning othello program. In 19th Annual Conference Gesellschaft für Klassifikation eV (Vol. 2).
• Liskowski, P., Jaskowski, W. M., & Krawiec, K. (2018). Learning to play Othello with deep neural networks. IEEE Transactions on Games.

License

GNU General Public License v3.0