登录 | 注册 | 充值 | 退出 | 公司首页 | 繁体中文 | 满意度调查
综合馆
基于遗传算法的Bayesian网结构学习研究
  • 摘要

    从不完备数据中学习网络结构是Bayesian网学习的难点之一,计算复杂度高,实现困难.针对该问题提出了一种进化算法.设计了结合数学期望的适应度函数,该函数利用进化过程中的最好Bayesian网把不完备数据转换成完备数据,从而大大简化了学习的复杂度,并保证算法能够向好的结构不断进化.此外,给出了网络结构的编码方案,设计了相应的遗传算子, 使得该算法能够收敛到全局最优的Bayesian网结构.模拟实验结果表明,该算法能有效地从不完备数据中学习.

  • 作者

    刘大有  王飞  卢奕南  薛万欣  王松昕 

  • 作者单位

    吉林大学计算机科学系

  • 刊期

    2001年8期 ISTIC EI PKU

  • 关键词

    Bayesian网学习  完备数据  不完备数据  计算复杂度  遗传算法 

参考文献
  • [1] Heckerman D. A tutorial on learning Bayesian network. Microsof t Research Tech Rep: MSR-TR-95-06, 1995
  • [2] 阎平凡;张长水. 人工神经网络与模拟进化计算. 北京:清华大学出版社, 2000
  • [3] Ashraf M;Abdelbar;Sandra M. Approximating MAPs for belief n etworks is NP-hard and other theorems. Artificial Intelligence, 1998,01
  • [4] Chickering D. Learning Bayesian networks is NP-complete. New York, 1996
  • [5] Larran ′aga P;Poza M;Yurramendi Y. St ructure learning of Bayesian networks by genetic algorithms: A performance analy sis of control parameters. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1996,09
  • [6] MAN LEUNG WONG;Kwong Sak Leung. Using evolutionary programming and min imum description length principle for data mining of Bayesian networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1999,02
  • [7] Beinlich I;Suermondt G. The ALRAM monitoring system: A case study with two probabilistic inference. Berlin, 1989
  • [8] Liu Dayou;Wang Fei. Learning Bayesian network(in Chinese). Haikou, 1999
  • [9] Yan Pingfan;Zhang Changshui. Artificial Neural Network and Simulating- Evolution Computation. 北京:清华大学出版社, 2000
  • [10] 刘大有;王飞. Bayesian网学习. 1999
  • [11] Friedman N;Heckerman D;Goldszmidt M. Challenge:Where is t he impact of Bayesian networks in learning?. Nagoya Japan, 1997
  • [12] Friedman N. The Bayesian structural EM algorithm. San Francisco, 1998
  • [13] Friedman N. Learning belief networks in the presence of missing values and hidden variables. Madison, 1997
  • [14] James W;Myers K;Laskey B;DeJong K A. Learning Bayesian n etworks from incomplete data using evolutionary algorithms. Stockholm Sweden, 1999
  • [15] Lauritzen S L. The EM algorithm for graphical association models with missing data. Computational Statistics and Data Analysis, 1995,02
  • [16] Binder J;Koller D;Russell S. Adaptive probabilistic networ ks with hidden variables. Machine Learning, 1997,02
  • [17] Madigan D;York J. Bayesian graphical models for discrete data. International Statistical Review, 1995,02
查看更多︾
相似文献 查看更多>>
18.210.22.132