面向决策行为预测的多层策略网络随机游走模型  

Multiplex strategy network random walk model for decision behavior prediction

作　　者：李晓寒 祁明泽 邓宏钟[1] LI Xiaohan;QI Mingze;DENG Hongzhong(College of Systems Engineering,National University of Defense Technology,Changsha 410073,China;College of Liberal Arts and Sciences,National University of Defense Technology,Changsha 410073,China)

机构地区：[1]国防科技大学系统工程学院,长沙410073 [2]国防科技大学文理学院,长沙410073

出　　处：《重庆理工大学学报（自然科学）》2022年第5期161-168,共8页Journal of Chongqing University of Technology：Natural Science

基　　金：国家自然科学基金项目(71771214,71690233,71971213);湖南省研究生科研创新项目(CX2020001)。

摘　　要：为了研究多阶段决策过程中人类决策行为的规律,提出了基于策略网络的决策行为建模与预测方法。通过单层策略网络的随机游走过程对多阶段决策过程进行建模;构建多层策略网络模型,利用网络层间游走和层内游走描述决策结果对决策行为的影响;在策略网络模型的基础上提出基于频率近似概率的单步策略预测算法,生成下一阶段的决策策略;通过多阶段决策仿真案例和典型博弈案例验证了方法的准确性和实用性。实验结果表明:提出的多层策略网络预测算法能够有效捕捉人的决策行为特征,在策略有明显规律性的情形下能达到70%的预测准确率。In order to study the regularity of human decision behavior in multi-stage decision making process,a method to model and predict decision behavior based on strategy network is proposed in the paper.Firstly,the multi-stage decision making process is characterized by the random walks of single-layer strategy network.Secondly,a multiplex strategy network is constructed to describe the influence of decision results on decision behavior by network inter-layer migration and intra-layer migration.Then,combined with frequency approximation probability,a single-step strategy prediction algorithm based on the strategy network model is proposed.Finally,through a multi-stage decision simulation process data and a typical game data,the accuracy and practicability of the above method are verified.Experimental results indicate that prediction algorithm multiplex strategy network based can effectively capture the characteristics of human decision behavior and reach 70%accuracy.

关 键 词：多层网络 随机游走 决策预测 

分 类 号：C934[经济管理—管理学]