A Bayesian Approach to Network Modularity

TL;DR

This paper introduces a Bayesian method for detecting network modules that accurately determines the number of modules and overcomes resolution limits, using variational techniques for model selection.

Contribution

It presents a novel Bayesian framework that unifies and extends existing module detection methods, with improved accuracy and interpretability.

Findings

Successfully recovers true number of modules in synthetic networks

Overcomes resolution limit problem in module detection

Applies effectively to real-world network data

Abstract

We present an efficient, principled, and interpretable technique for inferring module assignments and for identifying the optimal number of modules in a given network. We show how several existing methods for finding modules can be described as variant, special, or limiting cases of our work, and how the method overcomes the resolution limit problem, accurately recovering the true number of modules. Our approach is based on Bayesian methods for model selection which have been used with success for almost a century, implemented using a variational technique developed only in the past decade. We apply the technique to synthetic and real networks and outline how the method naturally allows selection among competing models.