基于结构分析的局部Gibbs抽样自动推理算法-模式识别与人工智能.PDFVIP

第26 卷摇 第4 期摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 模式识别与人工智能摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 Vol.26摇 No.4 摇 2013年4月摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 PR AI摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 Apr摇 摇 2013 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 摇 基于结构分析的局部Gibbs抽样自动推理算法* 王摇 浩摇 摇 曹龙雨摇 摇 姚宏亮摇 摇 李俊照 (合肥工业大学 计算机与信息学院摇 合肥230009) 摘摇 要摇 提出一种基于结构分析的局部Gibbs抽样的贝叶斯网络推理算法(S鄄LGSI).S鄄LGSI算法基于联合树算法 的概率图模型分析思想,对贝叶斯网络进行精确分解,然后根据查询结点和证据结点生成具有强相关性的局部网 络模型,进而对局部网络模型进行Gibbs抽样推理.与当前基于抽样的其它近似推理算法相比,该算法降低推理的 计算维数.同时,由于局部抽样模型包含了与查询结点相关的重要信息,因此该算法保证局部抽样推理的精度.算 法分析和在Alarm 网的实验结果表明,S鄄LGSI算法较显著降低时间复杂度,同时也提高推理精度.S鄄LGSI算法应 用于上海证券交易所股票网络的推理结果与实际情况基本一致,表现出较强的实用性. 关键词摇 自动推理,贝叶斯网络,马尔科夫蒙特卡洛,吉布斯抽样 中图法分类号摇 TP 181 A Local Gibbs Sampling Automatic Inference Algorithm Based on Structural Analysis WANG Hao,CAO Long鄄Yu ,YAO Hong鄄Liang ,LI Jun鄄Zhao (School of Computer and Information,Hefei University of Technology,Hefei23009) ABSTRACT In this paper, a local Gibbs sampling inference algorithm of Bayesian networks (S鄄LGSI) is proposed. Firstly,the S鄄LGSI algorithm precisely decomposes Bayesian networks based on the analytic idea ofjunction tree algorithm. Secondly,the suitable local model is chosen by the query node and the evidence node. Then,Gibbs sampling inference algorithmfor local network model isutilized. Compared with other current approximate sampling algorithms, the S鄄LGSI algorithm significantly reduces the calculation dimension. The sampling inference in the local model avoids the statistics of joint sample series and greatly reduces the calculation dimension. The proposed algorithm guarantees the inference precision,asthelocalmodelcontainsimportantinformationaboutthequerynode. Algorithman