Automatic Dimension Selection for a Non-negative Factorization Approach to Clustering Multiple Random Graphs

TL;DR

This paper introduces an automatic method for selecting the number of clusters when grouping multiple graphs using non-negative factorization and singular value thresholding, validated on real and simulated datasets.

Contribution

It proposes a novel model selection criterion for determining the number of clusters in graph clustering with non-negative factorization.

Findings

The method accurately estimates the number of clusters in various datasets.

It outperforms two standard clustering algorithms in experiments.

The approach is validated on both real and simulated data.

Abstract

We consider a problem of grouping multiple graphs into several clusters using singular value thesholding and non-negative factorization. We derive a model selection information criterion to estimate the number of clusters. We demonstrate our approach using "Swimmer data set" as well as simulated data set, and compare its performance with two standard clustering algorithms.