[1K3-GS-3] Knowledge utilization and sharing

[1K3-GS-3-03] A consideration on Othello program aiming for a draw based on Monte Carlo Tree Search

〇KENSHIRO TSUBOTA¹, Haruka Yamashita¹ (1. sophia university)

Keywords:Monte Carlo Tree Search, UCB1

There have been many cases of AIs defeating human world champions in perfect information games.However,since these AIs seek pure strength,it is difficult for inexperienced players to enjoy the game.Therefore,we propose an AI that matches the abilities of the opposing players and makes the game competitive.We propose an algorithm based on Monte Carlo Tree Search (MCTS).In previous Othello AI,the difficulty level was set based on the adjustment of hyper-parameters that increase the winning rate in relation to the objective function of the algorithm.This will create an AI in which matches are more likely to be draws,and is expected to create an AI that can be enjoyed by players of a wide range of levels.We will evaluate the validity of the proposed model by comparing the performance of the proposed model against each of the AI of a certain level.