J4

• 计算机科学 • 上一篇    下一篇

Bayesian网推理中的化简方法

董旭初, 欧阳丹彤, 刘大有   

  1. 吉林大学计算机科学与技术学院, 长春 130012
  • 收稿日期:2003-06-30 修回日期:1900-01-01 出版日期:2004-01-26 发布日期:2004-01-26
  • 通讯作者: 欧阳丹彤

Simplification on Bayesian network inference

DONG Xu-chu, OUYANG Dan-tong, LIU Da-you   

  1. College of Computer Science and Technology, Jilin University, Changchun 130012, China; Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education, Changchun 130012, China
  • Received:2003-06-30 Revised:1900-01-01 Online:2004-01-26 Published:2004-01-26
  • Contact: OUYANG Dan-tong

摘要: 针对Bayesian网的一般推理是NP难的, 在很多情况下, 具体查询时简化Bayesian网结构可以明显减少计算量的特点. 提出一种可以实现这种 策略的3个操作: 分离、 替换和遍历. 实验表明, 这种策略一般可以提高计算效率, 有时 会有明显作用.

关键词: Bayesian网, 分离操作, 替换操作, 遍历操作

Abstract: The inference in general Bayesian netw orks is NP hard. However, computations can be usually reduced by means of simpli fying the structures of Bayesian networks with respect to the given query. In th is paper, three operations, omitting, replacing and traversing operations, are p roposed in order to realize this strategy. Empirical tests revealed that this st rategy can improve computational efficiency generally, sometimes remarkably.

Key words: Bayesian network, omitting operation, replacing operat ion, traversing operation

中图分类号: 

  • TP301