A novel deep residual network-based incomplete information competition strategy for four-players Mahjong games

摘要

The game theory is widely acknowledged to benefit a lot from recent advances in deep learning, and intelligent competition strategies have been proposed for both complete information games and incomplete information games in recent years. In this paper, the four-players Chinese Mahjong game, which is a typical incomplete information game, is emphasized, a low-level semantic pseudo image generated based on game related prior knowledge and a novel deep residual network-based competition strategy are introduced to play the Chines Mahjong game. Technically, the deep learning within this new competition strategy is realized by a series of "GoBlock", which is a new deep learning model structure introduced in this paper. Also, the "GoBlock" is further made up of several "Inception+" sub-structures, which is novel as well. Comprehensive experiments are conducted to reveal the superiority of this new competition strategy. A great number of the Chinese Mahjong game data have been collected from an online Chinese Mahjong company to construct the dataset in this study, and the newly proposed competition strategy has been compared with several shallow learning-based methods as well as deep learning-based methods. Both qualitative and quantitative analysis have been conducted based on outcomes obtained by all compared methods, and the superiority of the new competition strategy over others are suggested. Furthermore, an interesting competition among the new AI competition strategy and three real senior players are also introduced in this paper. The effectiveness and efficiency of the new competition strategy over real senior human players are also revealed by quantitative analysis based on four measures, from the statistical point of view. It is also necessary to point out that, this work is the first attempt to tackle the Mahjong game, which is a typical incomplete information game, from the deep learning perspective.