Polygames: Improved Zero Learning
This work addresses the problem of scaling Zero learning to complex board games like Hex for AI researchers and developers, though it is incremental with architectural and training improvements.
The authors tackled the challenge of improving Zero learning for board games by developing Polygames, a framework that uses a fully convolutional architecture with global pooling to create board-size-independent bots and robust training via checkpoint tracking. They achieved victories against strong human players in Hex (19x19) and Havannah, and won first places in TAAI competitions.
Since DeepMind's AlphaZero, Zero learning quickly became the state-of-the-art method for many board games. It can be improved using a fully convolutional structure (no fully connected layer). Using such an architecture plus global pooling, we can create bots independent of the board size. The training can be made more robust by keeping track of the best checkpoints during the training and by training against them. Using these features, we release Polygames, our framework for Zero learning, with its library of games and its checkpoints. We won against strong humans at the game of Hex in 19x19, which was often said to be untractable for zero learning; and in Havannah. We also won several first places at the TAAI competitions.