J4

• 计算机科学 • 上一篇    下一篇

基于免疫进化算法的Bayesian网结构学习

于鹏, 刘大有, 贾海洋, 杨博   

  1. 吉林大学 计算机科学与技术学院, 长春 130012
  • 收稿日期:2006-01-12 修回日期:1900-01-01 出版日期:2006-11-26 发布日期:2006-11-26
  • 通讯作者: 刘大有

Learning Bayesian Network Structure Based on Immune Evolutionary Algorithms

YU Peng, LIU Dayou, JIA Haiyang, YANG Bo   

  1. College of Computer Science and Technology, Jilin University, Changchun 130012, China
  • Received:2006-01-12 Revised:1900-01-01 Online:2006-11-26 Published:2006-11-26
  • Contact: LIU Dayou

摘要: 利用免疫进化算法(IEA), 借助遗传和接种疫苗操作将基于打分和基于约束的两类Bayesian网结构学习方法有机地结合在一起, 提出一种新的Bayesian网结构学习方法. 通过与基于遗传算法的Bayesian网结构学习方法EGA(Expectation & Genetic Algorithm)的对比试验表明, 所提出算法的收敛速度更快、 学习得到网络的精度更高.

关键词: Bayesian网, 免疫进化算法, 遗传算法

Abstract: A new algorithm was put forward for learning Bayesian’s network structure. This algorithm is based on an improved Genetic Algorithm: Immune Evolutionary Algorithm (IEA). The IEA combines the two kinds of methods in learning’s Bayesian network, the score based method and the constraint based method, by genetic operation and vaccination operation. The experimental results show that the convergent speed of IEA is more rapid, and the precision of IEA is higher than that of EGA (Expectation & Genetic Algorithm, a method learning Bayesian network based on GA).

Key words: Bayesian network, immune evolutionary algorithm, genetic algorithm

中图分类号: 

  • TP301