数据缺失下学习贝叶斯网的SEM算法  

SEM Algorithm with Values Missed in Learning Bayesian Network

作　　者：廖学清[1] 吕强[1,2] 单冬冬[1] 

机构地区：[1]苏州大学计算机科学与技术学院,苏州215006 [2]江苏省计算机信息处理技术重点实验室,苏州215006

出　　处：《计算机工程》2009年第8期214-216,219,共4页Computer Engineering

基　　金：教育部博士点基金资助项目(20060285008);江苏省自然科学基金资助项目(BK2003030)

摘　　要：针对SEM算法易收敛于局部最优的问题,提出一种改进SEM算法——DSEM-PACOB算法,采用PACOB算法提供的良好候选网络及合理的策略,估计节点变量缺失值,并使与待估节点变量紧密相关的若干节点变量直接参与估计。实验结果表明,与SEM算法相比,利用新算法获得的最终解的质量有所提高,且具有更好的稳定性。Aiming at the problem that SEM algorithm always converges to local optimal network, an improved SEM algorithm called DSEM-PACOB algorithm is proposed, which makes use of the better candidate networks and reasonable strategy provided by PACOB algorithm to estimate the missing value of node variables. Meanwhile, it chooses several node variables, which have close correlations with the estimated node variable, to conduct estimation. Experimental results show that, compared with SEM algorithm, this new algorithm makes qualitative improvements on the quality of the final solutions, and has better performance of stability.

关 键 词：学习贝叶斯网 PACOB算法 紧密相关 新决策网络 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]