文献类型：期刊文献

中文题名：复杂网络社区发现的多目标五行环优化算法

英文题名：Multi-objective Five-elements Cycle Optimization Algorithm for Complex Network Community Discovery

作者：张清琪[1];刘漫丹[1]

机构：[1]华东理工大学信息科学与工程学院,上海200237

年份：2020

卷号：47

期号：8

起止页码：284

中文期刊名：计算机科学

外文期刊名：Computer Science

收录：CSTPCD;;北大核心:【北大核心2017】；CSCD:【CSCD_E2019_2020】；

语种：中文

中文关键词：复杂网络;多目标优化;社区发现;五行环优化算法;进化算法

外文关键词：Complex network;Multi-objective optimization;Community discovery;Five-elements cycle optimization;Evolutionary algorithm

摘要：社区结构作为复杂网络的重要特性,对理解网络的功能和结构具有重要意义。为了解决复杂网络的社区发现问题,提出了一种多目标五行环优化算法(Multi-Objective Five-Elements Cycle Optimization,MOFECO)。首先,将社区发现问题建模为多目标优化问题,选取反比率关联(Inverse Ratio Association,IRA)和比例缩减(Ratio Cut,RC)这两个互相对立的目标作为目标函数;然后,基于五行环模型(Five-Elements Cycle Model,FECM),通过局部最优解和全局最优解实现元素的更新,并引入交叉和变异算子对更新策略进行改进;最后,使用快速非支配排序的方法获得Pareto最优社区划分集合,有助于揭示复杂网络的层次结构。在人工合成网络和真实社会网络上进行实验,与单目标算法(GA-Net,Meme-Net)以及多目标算法(MOGA-Net,MOCD,MOEA/D-Net,DMOPSO,DIM-MOEA/D,MOCD-ACO)进行对比得出,MOFECO算法弥补了传统单目标优化社区结构划分单一的缺陷,在社区发现的准确度上有所提高。
As an important property of complex networks,community structure is of great significance for understanding the function and organization of networks.In order to solve the community discovery problem of complex networks,a multi-objective five-elements cycle optimization algorithm(MOFECO)is proposed.Firstly,the task of the community discovery is modelled as a multi-objective optimization problem,and two opposite targets,inverse ratio vssociation(IRA)and ratio cut(RC)are selected as the objective function.Then,based on five-elements cycle model(FECM),individual updating is implemented through local optimal individuals and global optimal individuals,and crossover and mutation operators are introduced to improve the update strategy.Finally,the fast non-dominated sorting method is used to obtain the Pareto optimal community partitioning set,which is helpful to reveal the hierarchical structure of complex networks.Experiments are conducted on LFR benchmark networks and real social networks,compared with single-objective algorithms(GA-Net,Meme-Net)and multi-objective algorithms,such as MOGA-Net,MOCD,MOEA/D-Net,DMOPSO,DIM-MOEA/D and MOCD-ACO,MOFECO overcomes the shortcomings of traditional single-objective optimization of community structure division,and improves the accuracy of community discovery.