核k-means聚类检测复杂网络社团算法

Kernel k-means Clustering Algorithm for Detecting Communities in Complex Networks

为揭示复杂系统中的结构与功能之间的联系,复杂网络中的社团发现成为一项最基本的任务。最近,李等人提出了一种用来评估社团质量的函数,称之为模块密度函数(即D值),并利用一个核矩阵给出了模块密度目标函数与核k-means方法之间的等价性。基于这种等价性,通过过渡操作的核矩阵来优化模块密度函数并提出了一种新的核k-means算法。实验结果表明,这种算法在发现复杂网络社团上是有效的。

Discovery community structure is fundamental for uncovering the links between structure and function in complex networks.In this context,recently,Li et al. recently proposed modularity density objective function for community detecting called the D function and gave the equivalence between modularity density objective function and the kernel k-means by using a kernel matrix.In this paper,based on this equivalence,we used the kernel matrix to optimize the modularity density and developed a new kernel k-means algorithm.Experimental results indicate that the new algorithms are efficient at finding community structures in complex networks.