• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊
论文

贝叶斯网络的无损分解

展开
  • (杭州电子科技大学通信工程学院,浙江 杭州 310018)
车璐(1985),男,河南新乡人,硕士生,研究方向为图像处理与传输;郭春生,博士,副教授,研究方向为图像分析与处理、软件无线电等。

收稿日期: 2009-02-04

  修回日期: 2009-05-18

  网络出版日期: 2010-03-28

Lossless Decomposition of the Bayesian Networks

Expand
  • (School of Communication Engineering,Hangzhou Dianzi University,Hangzhou 310018,China)

Received date: 2009-02-04

  Revised date: 2009-05-18

  Online published: 2010-03-28

摘要

处理复杂问题的途径和方法有很多,分而治之就是其中的一种有效方法。在将复杂问题分解为一些小问题的过程中,保存原始问题中的信息是关键。本文基于贝叶斯网络的联合树概念及其性质,提出了一种分解贝叶斯网络的方法,该方法可以有效地处理复杂的贝叶斯网络,并且能很好地解决分解过程中信息保存的问题。算法分解产生的各个小网络既保存了原始网络的依赖关系,又没有向分解产生的小网络增添新的依赖关系,因此该分解过程是无损的。最后借助典型的Asia网络详细地阐述了无损分解的整个过程,该例子也验证了无损分解方法的有效性。

本文引用格式

车璐,郭春生 . 贝叶斯网络的无损分解[J]. 计算机工程与科学, 2010 , 32(4) : 151 -153 . DOI: 10.3969/j.issn.1007130X.2010.

Abstract

Many ways and methods can be used to process the complex problems,such as divide and conquer,which is an important problemsolving technique. The key technology during the decomposition is the problem of information preservation. A method of decomposing a single Bayesian network is proposed based on the conception and properties of the Bayesian network junction tree, and it can process the complex Bayesian networks effectively, and can also solve the problem of information preservation well as decomposing a Bayesian network. Because no conditional independency information is lost and no extraneous conditional independency information is introduced during the decomposition, this method is lossless. Finally, the paper detailedly analyzes the process of decomposing a Bayesian network with the Asia Bayesian network, and the effectiveness of this method has been verified by this example.
文章导航

/