Cross-validation estimate of the number of clusters in a network

TL;DR

This paper introduces a new cross-validation based method for accurately estimating the number of clusters in network data, improving interpretability of modular structures.

Contribution

It presents a scalable, principled approach using leave-one-out cross-validation to assess the optimal number of clusters in networks.

Findings

Effective in various network types

Accurate estimation of cluster number

Scalable to large networks

Abstract

Network science investigates methodologies that summarise relational data to obtain better interpretability. Identifying modular structures is a fundamental task, and assessment of the coarse-grain level is its crucial step. Here, we propose principled, scalable, and widely applicable assessment criteria to determine the number of clusters in modular networks based on the leave-one-out cross-validation estimate of the edge prediction error.