A Study on Reinforcement Learning in Tri-Othello

The game of Tri-Othello is a newly designed three-player board game inspired by Othello. This paper modifies the rules of Tri-Othello to reduce the learning curve for human players while maintaining its fun and fairness. We implemented a game-playing program using reinforcement learning with the revised rules and compared its playing strength with a version based on the max<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</sup> search algorithm. Our experimental results showed that the trained reinforcement learning model performed better than the max<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</sup>-based program.