摘要
根据对叫牌和打牌阶段的研究,提出一种分步协同桥牌智能博弈策略。针对叫牌阶段叫品信息的关联性,设计了基于深度神经网络的叫牌模型,用于下一步叫牌决策,通过实验验证了模型的可行性。结合叫牌阶段的已知信息,提出由首攻策略、蒙特卡洛模拟、双明手求解算法构成的打牌策略。结果表明,该桥牌博弈策略可为不完全信息博弈研究提供参考。
Bridge is one of the most complex card games of significant research value.As a typical representative of incomplete information games,it hides more information and involves cooperation and competition between players.This paper proposes a step-by-step collaborative bridge intelligent game strategy based on the research of the two stages of calling and playing cards.A deep neural network-based bidding model is designed to address the correlation of bidding information during the bidding stage,and to provide the next bidding decision.The feasibility of the model is verified through experiments.A card playing strategy consisting of a first attack strategy,Monte Carlo simulation,and double hand solving algorithm is proposed based on the known information during the call stage.Experimental results indicate the bridge game strategy may provide some insights into the study of incomplete information games.
作者
王璐瑶
李学俊
吴蕾
WANG Luyao;LI Xuejun;WU Lei(College of Computer Science and Technology,Anhui University,Hefei 230601,China)
出处
《重庆理工大学学报(自然科学)》
北大核心
2025年第8期105-110,共6页
Journal of Chongqing University of Technology:Natural Science
关键词
定约桥牌
不完全信息
神经网络
蒙特卡洛模拟
contract bridge
imperfect information
neural network
Monte Carlo simulation
