基于信息流改进的贝叶斯网络结构学习算法

基于信息流提出贝叶斯网络结构学习的改进型搜索评分算法。首先计算信息流进行全局因果分析,构造0/1优化问题,获得最优初始网络结构;在此初始结构的基础上产生搜索空间,采用贪婪算法搜索最优结构弧,同时由信息流确定弧方向,实现网络结构的一体化学习。首次将信息流引入贝叶斯网络的结构学习,优化了初始搜索空间,实现了弧和弧方向的同步确定,更能获得近似全局最优结构。实验表明,改进算法较其他算法的准确性和学习效率更高。

Improved structure learning algorithm of Bayesian network based on information flow

An improved scoring search algorithm based on information flow is proposed for Bayesian network structure learning. Firstly, the 0/1 optimization problem is constructed based on the information flow for global causal analysis, and the optimal initial network structure is obtained. Then, the search space is generated based on the initial structure, and the optimal structure arcs are searched by the greedy algorithm. At the same time, the arc direction is determined by the information flow, to achieve integrated learning of the Bayesian network structure. For the first time, the information flow is introduced into the structure learning of Bayesian network, optimizing the initial search space, realizing the synchronous determination of arcs and arc direction, and obtaining the approximate global optimal structure. Experiments show that the improved algorithm has higher accuracy and learning efficiency than other algorithms.