高斯混合模型算法提取复杂网络社团  

Extracting Complex Network Community by Gauss Mixture Model Algorithm

作　　者：代婷婷[1] 董延寿[1] 韩艳[1] 陈洁[1] Dai Tingting Dong Yanshou Han Yan Chen Jie(School of Mathematics and Statistics, Zhaotong University, Zhaotong, Yunnan, 65700)

机构地区：[1]昭通学院数学与统计学院,云南昭通657000

出　　处：《保山学院学报》2017年第2期65-70,共6页JOURNAL OF BAOSHAN UNIVERSITY

基　　金：云南省应用基础研究项目(青年项目)(项目编号:2016FD082);昭通学院校级科学研究课题(项目编号:2016xj32)

摘　　要：基于复杂网络中的社团划分问题,提出了一种基于主成分分析的高斯混合模型社团提取算法。利用主成分分析对网络的邻接矩阵进行降维处理,假设一个网络中的社团由不同高斯模型生成,用期望最大化算法对模型的参数进行了求解。结果表明,当主成分的贡献率达90%以上时,网络的划分和实际网络非常吻合,所用时间也较短,表明该算法与以往方法相比具有明显优越性。In this paper, we propose a Gauss mixture model based on principal component analysis for community partition in complex networks. The idea of the algorithm is： firstly, the principal component analysis is used to reduce the dimension of the adjacency matrix; secondly, suppose that a community in a network is generated by different Gauss models, that is to say, the formation mechanism of different models is different; at last, the parameters of the model are solved by the expectation maximization algorithm. Simulation experiments show that if the contribution rate of the principal component is above ninety percent, the division of the network is in good agreement with the actual network, and the time is shorter. Compared with other methods, it has obvious superiority.

关 键 词：社团提取 主成分分析 高斯混合模型 EM算法 

分 类 号：O24[理学—计算数学]