【摘要】 马尔科夫网络中的隐藏变量学习是一个重要而困难的研究课题.关于隐藏变量需要解决三个问题,它们是隐藏变量的位置,维数和取值,局部结构.根据马尔科夫网络中的Clique确定隐藏变量的位置;然后基于依赖结构和Gibbssampling确定隐藏变量的取值和维数;最后利用MDL标准进行局部打分确定局部结构.试验结果表明,该方法能够有效地进行马尔科夫网络的隐藏变量学习.更多还原

【Abstract】 It is an important and difficult research project to learn hide variables in Markov network. Three problems should be solved when learning hide variable in Markov network. They are the location, dimension and values and local structure about hide variable. In this paper, They are respectively made sure by the clique in Markov network, combining dependency structure and Gibbs sampling and MDL criterion. Experimental results show that this method is effective in learning hide variables in Markov network.更多还原