Visual clustering of complex network based on nonlinear dimension reduction

Abstract

In this paper, we present a new visual clustering algorithm inspired by nonlinear dimension reduction technique: Isomap. The algorithm firstly defines a new graph distance between any two nodes in complex networks and then applies the distance matrix to Isomap and projects all nodes into a two dimensional plane. The experiments prove that the projected nodes emerge clear clustering property which is hidden in original complex networks and the distances between any two nodes reflect their close or distant relationships.