摘要
遗传算法是基于自然界中生物遗传规律的适应性原则对问题解空间进行搜寻和最优化的方法。贝叶斯网络是对不确定性知识进行建模、推理的主要方法,Bayesian网中的学习问题(参数学习与结构学习)是个NP-hard问题。强化学习是利用新顺序数据来更新学习结果的在线学习方法。介绍了利用强化学习指导遗传算法,实现对贝叶斯网结构进行有效学习。
Genetic algorithm is a kind of searching and optimization method which is based on the adaptation principles of the genetic laws found in nature. Bayesian networks are the main methods which are applied to conduction of modeling and reasoning for uncertainty knowledge, and the study problem over Bayesian networks is known to be NP-hard. Reinforcement learning is an online study method by using the new sequence data to update the study result. In this paper, we developed an efficient approach to the structur...
出处
《微型机与应用》
北大核心
2007年第S1期51-54,58,共5页
Microcomputer & Its Applications
基金
安徽省教育厅自然科学基金项目(KJ2007B152)资助
安徽省高校青年教师资助计划项目(2007jq1180)
皖西学院应用项目
关键词
贝叶斯网络
遗传算法
强化学习
Bayesian networks
genetic algorithm
reinforcement learning