基于免疫遗传算法的复杂网络社区发现
发布时间：2018-07-31 20:18
【摘要】：针对大部分基于智能优化算法的社区发现方法存在的种群退化、寻优能力不强、计算过程复杂、需要先验知识等问题,提出了一种基于免疫遗传算法(GA)的复杂网络社区发现方法。算法将改进的字符编码和相应的遗传算子相结合,在不需要先验知识的情况下可自动获得最优社区数和社区划分方案;将免疫原理引入遗传算法的选择操作中,保持了群体多样性,改善了遗传算法所固有的退化现象;在初始化种群及交叉和变异算子中利用网络拓扑结构的局部信息,有效缩小了搜索空间,增强了寻优能力。计算机生成网络和真实网络上的仿真实验结果表明算法可自动获取最优社区数和社区划分方案并具有较高的精度,说明算法具有可行性和有效性。
[Abstract]:In order to solve the problems of population degradation, poor optimization ability, complex calculation process and the need of prior knowledge in most community discovery methods based on intelligent optimization algorithms, etc. A complex network community discovery method based on immune genetic algorithm (GA) is proposed. By combining the improved character encoding with the corresponding genetic operator, the optimal community number and community partition scheme can be obtained automatically without prior knowledge, and the immune principle is introduced into the selection operation of genetic algorithm. The diversity of population is preserved and the inherent degradation of genetic algorithm is improved. The local information of network topology is used in initializing population and crossover and mutation operator to effectively reduce the search space and enhance the searching ability. The simulation results on the computer generated network and real network show that the algorithm can automatically obtain the optimal community number and community partition scheme, and has a high accuracy, which shows that the algorithm is feasible and effective.
【作者单位】： 西北民族大学数学与计算机科学学院;
【基金】：国家自然科学基金资助项目(11161041) 2012年度国家民委科研项目基金资助项目 中央高校基本科研项目基金资助项目(31920130009,zyz2012081)
【分类号】：TP18;TP393.02
[Abstract]:In order to solve the problems of population degradation, poor optimization ability, complex calculation process and the need of prior knowledge in most community discovery methods based on intelligent optimization algorithms, etc. A complex network community discovery method based on immune genetic algorithm (GA) is proposed. By combining the improved character encoding with the corresponding genetic operator, the optimal community number and community partition scheme can be obtained automatically without prior knowledge, and the immune principle is introduced into the selection operation of genetic algorithm. The diversity of population is preserved and the inherent degradation of genetic algorithm is improved. The local information of network topology is used in initializing population and crossover and mutation operator to effectively reduce the search space and enhance the searching ability. The simulation results on the computer generated network and real network show that the algorithm can automatically obtain the optimal community number and community partition scheme, and has a high accuracy, which shows that the algorithm is feasible and effective.
【作者单位】： 西北民族大学数学与计算机科学学院;
【基金】：国家自然科学基金资助项目(11161041) 2012年度国家民委科研项目基金资助项目 中央高校基本科研项目基金资助项目(31920130009,zyz2012081)
【分类号】：TP18;TP393.02
【参考文献】
相关期刊论文 前4条
1 周世兵;徐振源;唐旭清;;新的K-均值算法最佳聚类数确定方法[J];计算机工程与应用;2010年16期
2 罗锦坤;元昌安;杨文;胡卉颖;袁晖;;基于基因表达式编程算法的复杂网络社区结构划分[J];计算机应用;2012年02期
3 何东晓;周栩;王佐;周春光;王U，
本文编号：2156720
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2156720.html
