Exact integrated completed likelihood maximisation in a stochastic block transition model for dynamic networks

TL;DR

This paper introduces a scalable Bayesian method for fitting stochastic block transition models to dynamic networks, effectively capturing edge persistence and automatically determining the optimal number of groups.

Contribution

It presents a greedy optimization algorithm for exact integrated completed likelihood maximization in stochastic block transition models, enabling scalable and automatic model selection.

Findings

Efficient algorithm scales to large datasets

Automatically determines the number of latent groups

Proven effective on artificial and real network data

Abstract

The latent stochastic block model is a flexible and widely used statistical model for the analysis of network data. Extensions of this model to a dynamic context often fail to capture the persistence of edges in contiguous network snapshots. The recently introduced stochastic block transition model addresses precisely this issue, by modelling the probabilities of creating a new edge and of maintaining an edge over time. Using a model-based clustering approach, this paper illustrates a methodology to fit stochastic block transition models under a Bayesian framework. The method relies on a greedy optimisation procedure to maximise the exact integrated completed likelihood. The computational efficiency of the algorithm used makes the methodology scalable and appropriate for the analysis of large network datasets. Crucially, the optimal number of latent groups is automatically selected at no additional computing cost. The efficacy of the method is demonstrated through applications to both artificial and real datasets.