Learning Bayesian Network Structure Based on Bernoulli Distribution
基于贝努里分布的贝叶斯网络结构学习算法

At present,the algorithm of learning bayesian structure with missing data is mainly based on the search and scoring method combined with EM algorithm.The algorithm has low efficiency.In this paper,a new algorithm of learning Bayesian network structure with missing data is presented.First,we adopt the Bernoulli distribution to express the relationship between the variables in database.Second,we use KL divergence to express the similarity between the cases.Third,we draw the value of the missing data according to the Gibbs sampling.Finally,we use heuristical search to complete the learning of Bayesian network structure.This method can avoid the exponential complexity of standard Gibbs sampling and the main problems in the existing algorithm.