Improvement Of Monte Carlo Tree Search Algorithm In Two-person Game Problem

Artificial intelligence is now very important research areas, not only in the computer field, all walks of life have a wide range of applications. Machine learning is an important branch of artificial intelligence, with the continuous development of machine learning methods, people understand the artificial intelligence also has a deeper understanding, from the guidance of computer learning logic reasoning, to the church computer some prior knowledge made expert system, And now let the computer learn to self-study. Not only in dealing with large data, artificial intelligence has a wide range of use, in guiding human development strategy also has a more important guiding role.The AI algorithm in the double game game is the important development direction and application prospect of artificial intelligence. The emergence of AlphaGo marks the biggest problem on the double game game Go is also broken, AlphaGo cleverly to the depth of learning and Monte Carlo tree search algorithm, convolution neural network and other methods together, greatly enhance the Go AI calculation efficiency , So that under the human rule, the computer beat the best professional players to become a reality. AlphaGo success does not mean that the current algorithm is the best, in the study process found that Monte Carlo tree search algorithm there are still a lot of problems and hidden dangers.Monte Carlo tree search (MCTS) is a heuristic search algorithm for some kinds of decision processes, most notably those employed in game play. In the decision process of a very complex game like a computer Go game, the basic Monte Carlo tree search method converges very slowly due to large computation cost. In this article, we have indicated that MCTS cannot converge to the best strategy of the two-person complete information games. Therefore, we propose a new search algorithm which combines MCTS with the Min-Max algorithm in order to avoid the failure due to randomness of Monte Carlo method. For further enhancement of computation efficiency of MTCS in complex two-person games we also consider to employ some progressive pruning strategies. An experimental test has shown that the new algorithm significantly improves the accuracy and efficiency of MCTS.