Community detection in networks via nonlinear modularity eigenvectors
Abstract
Revealing a community structure in a network or dataset is a central problem arising in many scientific areas. The modularity function is an established measure quantifying the quality of a community, being identified as a set of nodes having high modularity. In our terminology, a set of nodes with positive modularity is called a \textit{module} and a set that maximizes is thus called \textit{leading module}. Finding a leading module in a network is an important task, however the dimension of real-world problems makes the maximization of unfeasible. This poses the need of approximation techniques which are typically based on a linear relaxation of , induced by the spectrum of the modularity matrix . In this work we propose a nonlinear relaxation which is instead based on the spectrum of a nonlinear modularity operator . We show that extremal eigenvalues of provide an exact relaxation of the modularity measure , however at the price of being more challenging to be computed than those of . Thus we extend the work made on nonlinear Laplacians, by proposing a computational scheme, named \textit{generalized RatioDCA}, to address such extremal eigenvalues. We show monotonic ascent and convergence of the method. We finally apply the new method to several synthetic and real-world data sets, showing both effectiveness of the model and performance of the method.
Cite
@article{arxiv.1708.05569,
title = {Community detection in networks via nonlinear modularity eigenvectors},
author = {Francesco Tudisco and Pedro Mercado and Matthias Hein},
journal= {arXiv preprint arXiv:1708.05569},
year = {2018}
}