Bayesian model-based clustering for multiple network data

Abstract

There is increasing appetite for analysing multiple network data. This is different to analysing traditional data sets, where now each observation in the data comprises a network. Recent technological advancements have allowed the collection of this type of data in a range of different applications. This has inspired researchers to develop statistical models that most accurately describe the probabilistic mechanism that generates a network population and use this to make inferences about the underlying structure of the network data. Only a few studies developed to date consider the heterogeneity that can exist in a network population. We propose a Mixture of Measurement Error Models for identifying clusters of networks in a network population, with respect to similarities detected in the connectivity patterns among the networks' nodes. Extensive simulation studies show our model performs well in both clustering multiple network data and inferring the model parameters. We further apply our model on two real-world multiple network data sets resulting from the fields of Computing (Human Tracking Systems) and Neuroscience.

Modelling Populations of Interaction Networks via Distance Metrics

“…Le et al (2018), Peixoto (2018 and Newman (2018) propose measurement error models which view observed networks as noisy realisations of an unknown ground truth. Along similar lines, Mantziou et al (2021) and Young et al (2022) have recently extended the measurement error models to capture heterogeneity, providing model-based approaches to clustering networks.…”

Clustering of heterogeneous populations of networks

“…After this work was completed we learned of very recent related work by Mantziou et al A preprint of their work can be found at [40].…”

Compressing network populations with modal networks reveals structural diversity

“…Other bodies of work have tailored populations of networks, including measurement approaches [4,7,15] and models that generalize the concept of unimodal distributions to networks [11,14]. Several recent contributions have directly addressed the problem of summarizing populations of networks when multiple modal networks are needed [9,17,18,[23][24][25][26][27][28][29]. All these approaches require specifying the number of modes ahead of time.…”

“…All these approaches require specifying the number of modes ahead of time. Current solutions to this problem include scree plots [9], regularization with ad hoc priors [18], over-parametrized models [24,25,28], or approximative information criteria [17,23] poorly adapted to network problems. These approaches also demand extensive tuning and significant computational overhead from fitting the model to several choices of the number of modes.…”