摘要
本文旨在研究如何将强化学习模型合理地应用在海克斯棋博弈算法中,并给出程序实现方案。以蒙特卡洛树搜索生成数据集训练卷积神经网络的方式,使得模型能够在不断自我对弈的过程中,修正自身选择动作的策略,更新模型参数,从而达到提升棋力的目的。实验结果表明,通过强化学习算法能够准确地评估海克斯棋的局面,并有效地选择有利的落子位置,使得海克斯棋博弈系统获得高质量的决策能力。
The purpose of this paper is to study how to apply reinforcement learning model to the algorithm of Hex game reasonably,and give the program implementation scheme.In this way,the convolution neural network can be trained by using the data set generated by the Monte Carlo tree search,so that the model can enhance chess skills by modifying the strategy of its own choice of action and updating the model parameters in the process of continuous self playing.The experimental results show that the reinforcement learning algorithm can accurately evaluate the situation of Hex game,and effectively select a favorable moves,so that Hex game system gains high-quality decision-making ability.
作者
张芃芃
孟坤
杨震栋
ZHANG Pengpeng;MENG Kun;YANG Zhendong(Computer School,Beijing Information Science&Technology University,Beijing 100192,China)
出处
《智能计算机与应用》
2020年第3期142-145,共4页
Intelligent Computer and Applications
基金
北京信息科技大学2019年促进高校内涵发展-大学生科研训练项目(5101923400)。