机构地区: 华南理工大学计算机科学与工程学院
出 处: 《控制理论与应用》 2011年第9期1135-1140,共6页
摘 要: 从优化模块度的角度出发,提出了一种基于粒子群优化的网络社区发现的粒子群优化算法(CDPSO);该算法根据网络连接数据的特点给出一种新的粒子编码方法,有效地避免非法粒子的产生,一定程度上缓解了基于二值编码的迭代二划分策略所遭遇的局部最优划分问题,并改进了传统离散粒子群优化(PSO)的粒子位置调整策略,使算法收敛速度更快.实验结果表明,CDPSO能够在无先验信息的条件下快速有效地揭示网络内在的社区结构. For optimizing the modularity,a community discovery algorithm(CDPSO) is proposed based on particle-swarm-optimization(PSO).By the characteristics of network link data,a novel particle-encoding scheme is presented to avoid the production of illegal particles,alleviate the local optimal-partition encountered in the iterative partition approach based on Boolean encoding scheme,and improve the particle-position adjustment strategy in traditional discrete PSO to achieve better convergence.Experimental results show that CDPSO can rapidly and effectively discover the intrinsic community structure in networks without any domain information.
领 域: [自动化与计算机技术] [自动化与计算机技术]