A0618
Title: Flexible modelling of heterogeneous populations of networks: A Bayesian nonparametric approach
Authors: Francesco Barile - Bicocca University (Italy) [presenting]
Bernardo Nipoti - University of Milan Bicocca (Italy)
Simon Lunagomez - Lancaster University (United Kingdom)
Abstract: The increasing availability of multiple network data has been calling for the development of statistical models for heterogeneous populations of networks. A popular approach to the problem of clustering multiple network data uses distance metrics that measure the similarity among networks based on some of their global or local characteristics. In this context, a novel Bayesian nonparametric approach is proposed to model undirected labelled graphs sharing the same set of vertices, which allows us to identify clusters of networks characterized by similar patterns in the connectivity of nodes. Our construction relies on the definition of a location-scale Dirichlet process mixture of centred Erdos-Renyi (CER) kernels. A unique mode or network representative and a univariate measure of dispersion around the mode conveniently parametrize the CER kernel function. An efficient Markov chain Monte Carlo scheme is proposed to carry out posterior inference and conveniently cluster the multiple network data. The number of clusters in the population is not set a priori but inferred from the data. The performance of our approach is investigated by means of an extensive simulation study and illustrated with the analysis of a dataset on brain networks.
