Learning to Play Othello with Deep Neural Networks

TL;DR

This paper investigates the effectiveness of convolutional neural networks for move prediction and playing strength in Othello, demonstrating they outperform existing methods and even surpass top open-source players.

Contribution

It is the first comprehensive evaluation of CNN architectures for Othello, showing their high predictive accuracy and strong playing performance, including defeating top open-source programs.

Findings

CNNs outperform previous 1-ply Othello players.

Best CNNs beat 2-ply Edax, a top open-source player.

High correlation between move prediction accuracy and playing strength.

Abstract

Achieving superhuman playing level by AlphaGo corroborated the capabilities of convolutional neural architectures (CNNs) for capturing complex spatial patterns. This result was to a great extent due to several analogies between Go board states and 2D images CNNs have been designed for, in particular translational invariance and a relatively large board. In this paper, we verify whether CNN-based move predictors prove effective for Othello, a game with significantly different characteristics, including a much smaller board size and complete lack of translational invariance. We compare several CNN architectures and board encodings, augment them with state-of-the-art extensions, train on an extensive database of experts' moves, and examine them with respect to move prediction accuracy and playing strength. The empirical evaluation confirms high capabilities of neural move predictors and suggests a strong correlation between prediction accuracy and playing strength. The best CNNs not only surpass all other 1-ply Othello players proposed to date but defeat (2-ply) Edax, the best open-source Othello player.