基于信息熵的Markov网络结构学习算法研究

Algorithm research of Markov network structural learning based on information entropy

Markov网是一种进行不确定性推理的有力工具,是一个无向图,不需要发现边的方向。基于信息熵中测试信息独立理论,对Markov网中各节点进行条件独立(CI)测试,并且基于信息熵的Markov网,提出一个有效的构造算法,大大提高推理效率,对以后知识的不确定推理研究具有一定的参考价值。

Markov network is a forceful tool to practice inference of uncertainty.It is an undirected graph,and it doesn＇t need to find the direction of an edge.On the basis of testing information independence theory,the test of C1 is carried out on all the nodes in the Markov network.Meanwhile,an effective structural algorithm based on information entropy is pre- sented in this paper,it greatly improves the reasoning efficiency, and has the reference value for uncertain reasoning re- search of knowledge.