基于优化蚁群和标签传播的复杂网络社区检测  

Community detection for complex networks based on optimization of ant colony and tag propagation

作　　者：付立东[1,2] 郭亚鑫[1] 宋进福 FU Li-dong;GUO Ya-xin;SONG Jin-fu(College of Computer Science and Technology,Xi’an University of Science and Technology,Xi’an 710699,China;School of Computer Science and Technology,Xidian University,Xi’an 710126,China)

机构地区：[1]西安科技大学计算机科学与技术学院,陕西西安710699 [2]西安电子科技大学计算机科学与技术学院,陕西西安710126

出　　处：《计算机工程与设计》2023年第5期1320-1327,共8页Computer Engineering and Design

基　　金：国家自然科学基金面上基金项目(61772394);陕西省自然科学基础研究计划基金项目(2020JM-526)。

摘　　要：标签传播算法被广泛应用于复杂网络中社区检测及其它工程领域,但由于其标签更新的随机性,降低了社区检测的稳定性,为此提出一种LPA-5SA(LPA-five step Ant)算法。使用蚁群优化算法的概率转移公式将标签传播算法的随机选择变为目标函数高概率选择,通过5步更新法提高信息素选择权重,提高社区检测的稳定性和准确率。该算法在真实的网络和人工合成的网络中进行实验,结果用模块度和NMI指标进行评价,验证了该方法的准确率和稳定性。Label propagation algorithm is widely used in community detection and other engineering fields in complex networks.However,due to the randomness of tag updating,the stability of community detection is reduced.A LPA-five step ant(LPA-5SA)was proposed.The probability transfer formula of ant colony optimization(ACO)algorithm was used to change random selection of label propagation algorithm(LPA)into high probability selection of objective function,and the pheromone selection weight was increased using five-step updating method,which greatly improved the stability and accuracy of community detection.The algorithm was tested in real networks and artificial networks,module degree and normalized mutual information(NMI)were used for evaluation.The accuracy and stability of the method are verified.

关 键 词：社区检测 蚁群算法 模块度 标签传播算法 5步蚁群算法 复杂网络 标准化互信息 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]