Title: Bayesian modeling of multiple network data Authors:  Francesco Barile - Bicocca University (Italy)
Simon Lunagomez - ITAM Mexico City (Mexico)
Bernardo Nipoti - University of Milan Bicocca (Italy) [presenting]
Abstract: A flexible framework is discussed for modeling multiple network data using similarity metrics to compare networks. Within this setting, we introduce a novel Bayesian nonparametric model that identifies clusters of networks with similar connectivity patterns. Our approach is based on a location-scale Dirichlet process mixture of centered Erdos-Renyi kernels, where each component is defined by a representative network, or mode, and a univariate measure of dispersion around it. This model offers desirable properties, including full support in the Kullback-Leibler sense and strong consistency. For posterior inference and network clustering, we develop an efficient Markov chain Monte Carlo algorithm. The models performance is evaluated through extensive simulations and applications to human brain network data and a global food trade dataset.