Dimension reduction using spectral methods in FANNY for fuzzy clustering of graphs

Abstract

FANNY is a fuzzy or soft clustering algorithm, where each node in the graph is associated with a membership coefficient, indicating degree of belongingness of each node to different clusters. In this paper, we proposed a method for multiple dimension reduction of feature space of graphs or networks by using Spectral methods for FANNY clustering algorithm. Simulations of our method on two real networks show that, the proposed algorithm produced better result than traditional FANNY in-terms of runtime as well as modularity.