基于l1惩罚加速的贝叶斯网贪婪搜索算法

Accelerating GDS algorithm with penalty for Bayesian network

首先通过l1惩罚学习高斯图模型(即贝叶斯网的道义图),然后在道义图上贪婪搜索贝叶斯网的结构。通过模拟实验与PC算法、GES方法和GDS算法进行了比较。

With a Gaussian graphical model(i.e.,the moral graph of Bayesian network)by l1 penalty,greedy searching structure of Bayesian network is realized based on the moral graph.Through simulation,the method is compared with other ones such as PC algorithm,GES and GDS.