摘要
给出了一种利用已知数据样本集,自动构造因果图的结构的学习算法,该算法以最小描述长度评分函数(MDL)为评价准则,这种学习算法对数据不需要任何先验假设,可以学习任意结构的因果图,同时兼顾学习出的因果图结构的精确性和复杂性,实验结果显示了算法的有效性和可行性.
A learning algorithm of constructing the Causality Diagram structure automatically according to a set of data is presented. The algorithm use minimum description length scoring function as evaluate rule. The algorithm does not need any prior hypothesis for the data set, it can learn all kind of Causality Diagram structure, and it can take the consideration of the Causality Diagram structure＇ s accuracy as well as its complexity. Experimental results show the validity and the feasibility of the new algorithm.
出处
《微电子学与计算机》
CSCD
北大核心
2009年第7期29-31,共3页
Microelectronics & Computer
基金
重庆市教委科学技术研究项目(KJ080817)
运筹学与系统工程重庆市市级重点实验室
重庆高校创新团队建设计划资助项目
