Recommending Bids on Dou-DiZhu Poker Games: A Deep Learning Approach
Bo Yuan, Shuqin Li · 2020
Computer aided game playing strategy provides a unique view towards artificial intelligence studies. In addition to symmetric games, establishing intelligent models for non-symmetric games are more challenging. A vivid example is the Dou-DiZhu game, a famous traditional Chinese poker game, which requires one player playing cards against the alliance of the other two players. This paper has made a first attempt to establish a machine-aided playing strategy to help player to call bids with a higher gain. In particular, a Convolutional Neural Network (CNN) based recommendation model is proposed to combine the feature of player's hand sequence and strength. To validate the feasibility of the proposed method, a real-world dataset is utilized to examine the performance of the recommendation results outputted the proposed model. The results of comparative experiments show that the proposed model achieves a higher precision, recall, F1 and accuracy performance than traditional machine learning models and single-feature based deep learning models.