首页 | 本学科首页   官方微博 | 高级检索  
     

采用约束蚁群优化的贝叶斯网结构学习算法
引用本文:王翔,郑建国,张超群,阎瑞霞. 采用约束蚁群优化的贝叶斯网结构学习算法[J]. 西安交通大学学报, 2011, 45(8): 54-61
作者姓名:王翔  郑建国  张超群  阎瑞霞
作者单位:东华大学旭日工商管理学院,200092,上海
基金项目:国家自然科学基金资助项目(70971020)
摘    要:针对贝叶斯网络结构学习问题,提出了一种约束蚁群优化算法.该算法根据贝叶斯得分的局部一致性原理设计了一种增边规则,并且将此规则引入蚁群算法的框架中,从而实现了在搜索过程中利用启发式信息动态缩减搜索空间、同时减少运行时间的目的.此外,还从理论上证明了增边规则的正确性,而且从实验角度讨论了约束蚁群优化算法的参数敏感性.实验结果表明,在解决较大规模的贝叶斯网络结构学习问题时,约束蚁群优化算法在保证求解精度的条件下比蚁群优化算法的运行时间减少40%以上.

关 键 词:贝叶斯网络  约束蚁群优化算法  增边规则

A Constrained Ant Colony Optimization Algorithm for Learning Bayesian Networks
WANG Xiang,ZHENG Jianguo,ZHANG Chaoqun,YAN Ruixia. A Constrained Ant Colony Optimization Algorithm for Learning Bayesian Networks[J]. Journal of Xi'an Jiaotong University, 2011, 45(8): 54-61
Authors:WANG Xiang  ZHENG Jianguo  ZHANG Chaoqun  YAN Ruixia
Affiliation:WANG Xiang,ZHENG Jianguo,ZHANG Chaoqun,YAN Ruixia(Glorious Sun School of Business and Management,Donghua University,Shanghai 200092,China)
Abstract:A novel constrained ant colony optimization algorithm is proposed for learning Bayesian networks.An add-edge-rule is designed in the proposed algorithm based on the locally consistent scoring criterion of BDEu metric.The add-edge-rule is embedded into the frame of ant colony optimization so that the heuristic information can be dynamically used to restrict the candidate evaluations during the search process and the running time of the new algorithm can be reduced.In addition,the correctness of the add-edge-...
Keywords:Bayesian networks  constrained ant colony optimization algorithm  add-edge-rule  
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号