摘要
由于玩家在对局中会有不同的博弈风格,针对竞技二打一中玩家叫牌风格划分的问题,提出了通过计算玩家叫牌和基准叫牌之间差异的统计学方法,设计了融合叫牌情况和玩家手牌及强力牌特征的基准叫牌模型来给出基准叫牌。实验结果表明:该方法中的基准叫牌模型能够在测试集上准确率达到94%以上,在玩家的叫牌风格实验中达到了预期的划分效果,在一定程度上验证了玩家博弈风格划分的可行性。
As players have different game styles in the game,aiming at the type of players’bidding style in Fighting the Landlord,this paper presents a statistical method to calculate the difference between player bidding and benchmark bidding,and designs a benchmark bidding model that combines the bidding situation with players’hand cards and dominating cards to make benchmark bidding.The results show that the benchmark bidding model in this method can achieve an accuracy rate of more than 94%on the test set,and achieve the expected division effect in players’bidding style experiments.To some extent,it verifies the feasibility of the division of players’game style.
作者
赵煜霖
沈强望
李淑琴
孟坤
ZHAO Yulin;SHEN Qiangwang;LI Shuqin;MENG Kun(Computer School,Beijing Information Science and Technology University,Beijing 100101,China;Perception and Computation Intelligence Joint Lab,Beijing 100101,China)
出处
《重庆理工大学学报（自然科学）》
CAS
北大核心
2022年第12期155-161,共7页
Journal of Chongqing University of Technology：Natural Science
基金
北京信息科技大学科技项目(5212010937,KM201911232002)。
关键词
博弈算法
竞技二打一
叫牌风格
卷积神经网络
game algorithm
Fighting the Landlord
bidding style
convolutional neural networks
