Position Control and Production of Various Strategies for Game of Go Using Deep Learning Methods
Authors: Fan, T., Shi, Y., Li, W., Ikeda, K.
Journal: International Journal of Information Science and Engineering (IJISE), ISSN 1694-4496
Citation: IJISE 4(1), 2022-08-13.
DOI: 10.1109/taai48200.2019.8959895
Type: Original Research
Abstract
This paper proposes a new control strategy for the game of Go, which combines deep learning methods with traditional game theory approaches. The proposed method utilizes a deep convolutional neural network to predict the best next move, and then uses a Monte Carlo Tree Search to explore the game tree and refine the move selection. Experimental results show that the proposed method outperforms existing Go AI programs in terms of winning rate and playing strength. The paper also discusses the implications of this new strategy for the development of AI in other complex games.
Keywords
Go AI, Deep Learning, Monte Carlo Tree Search, Game Theory, Artificial Intelligence
Full Text
This paper proposes a new control strategy for the game of Go, which combines deep learning methods with traditional game theory approaches. The proposed method utilizes a deep convolutional neural network to predict the best next move, and then uses a Monte Carlo Tree Search to explore the game tree and refine the move selection. Experimental results show that the proposed method outperforms existing Go AI programs in terms of winning rate and playing strength. The paper also discusses the implications of this new strategy for the development of AI in other complex games.