贝叶斯网络精确推理算法的研究

贝叶斯网络是以概率理论为基础的不确定知识表示模型,贝叶斯网络推理的目的是得到随机变量的概率分布。目前,最流行的推理算法是联合树算法,它的主要思想是将贝叶斯网络转化为一棵无向树,在无向树上完成消息传递过程,求出原贝叶斯网络中任意随机变量的概率分布。为了降低算法的计算时空复杂度,对算法进行了不断的改进,为贝叶斯网络推理算法的进一步研究提供了条件。

Research of exact inference algorithm in Bayesian Networks

Bayesian Network is an uncertain model based on Probability .The purpose of inference is to acquire posterior probability of each variable. At present, Junction tree arithmetic is the most popular .It translates Bayesian Network into a junction tree and in the tree messages are propagated, and then probability of every variable can come out. In order to reduce the complexity, we improve the arithmetic, which offers condition for studying inference arithmetic deep.