复杂网络社区的抽样概率分布估计检测算法  

SPEDAs:Sampling Probability Distribution Estimation Algorithm for Complex Network Community

作　　者：林洋[1] 李燕[1] 董玮[1] 刘延昕[1] 任丽晔[2] 

机构地区：[1]吉林广播电视大学远程教育技术中心,长春130022 [2]长春大学电子信息工程学院,长春130022

出　　处：《西南师范大学学报（自然科学版）》2016年第10期96-103,共8页Journal of Southwest China Normal University(Natural Science Edition)

基　　金：吉林省教育厅自然科学基金项目(2012243);吉林省科技厅自然科学基金项目(201215111)

摘　　要：针对复杂网络密集区域(社区)存在大量节点及稀疏网络,传统方法性能无法满足检测指标要求的问题,提出一种复杂网络社区的抽样概率分布估计检测算法.首先,针对社区检测方法中数学模型不够精确导致正确发现社区数量不足的问题,基于模块化密度(D值)提出一种无需先验知识的混合整数非线性社区检测模型;其次,针对该模型优化的NP难问题,采用Gibbs抽样概率模型,提高分布估计种群的普适性,并为优秀个体构造抽样学习样本,提高算法的优化性能;最后,通过在标准测试函数及复杂网络社区检测应用中验证了所提算法的有效性.Due to the problem of existence of a large number of nodes and sparse networks in the complex network intensive regional（community）,and the performance of traditional method could not meet the indexes demand,the sampling probability distribution estimation algorithm for complex network community has been proposed in this paper.Firstly,in order to solve the problem of insufficient quantity in finding the right model of community because of the inaccurate method,the community detection model has been proposed without prior knowledge based on the modularity density（Dvalue）.Secondly,according to the problem of NP-hard optimization,the Gibbs sampling probability model has been proposed to improve the universal of distribution estimation population,which could structure the learning sample for outstanding individuals,and the optimization performance of the algorithm been improved.Finally,the validity of proposed algorithm is verified by the standard test functions and complex network community detection application.

关 键 词：复杂网络 社区 抽样概率 分布估计 进化计算 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]