A comparison of machine leaming methods using a two player board game

The board games are usually performed by game theory algorithms: minimax and minimax with alpha-beta pruning. Tic-tac-toe (X-O) is the best-known two-player board game. The game tic-tac-toe, based on machine learning algorithms, has been shown in this research. The neural network has been developed and trained to play the game utilizing three implemented agents: an agent based on deep Q-learning, an agent based on policy gradient method and a random agent. The agent can play the game perfectly in the 10-minute training interval, on an average graphics processing unit.