Automated Detection of Edge Clusters via an Overfitted Mixture Prior

• Activity Number: 143 - SPEED: Bayesian Methods and Social Statistics Part 1
• Type: Contributed
• Date/Time: Monday, August 8, 2022 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Hanh Pham* and Daniel Sewell
• Companies: University of Iowa and University of Iowa
• Keywords: clustering; community detection; network analysis; overfitted mixture model; variational Bayes

Abstract

Most community detection methods focus on clustering actors that share some common features in a network. However, clustering edges offer a more intuitive way to understand the network structure in many real-life applications. Out of a few existing methods for network edge clustering, most are algorithmic, except for the latent space edge clustering (LSEC) model proposed by Sewell, 2020. LSEC was shown to have good performance in simulation and real-life data analysis, but fitting this model requires advance knowledge of the number of clusters and latent dimensions, which are often unknown to researchers. Within a Bayesian framework, we propose an extension to the LSEC model using a sparse mixture prior that supports automated selection of the number of clusters. Estimations for our automated LSEC (aLSEC) model are obtained efficiently via a variational Bayes generalized expectation-maximization approach. Our simulation study showed that aLSEC reduced run time by 10 to over 100 times compared to LSEC. Like LSEC, aLSEC has a computational cost that grows linearly with the number of actors in a network, making it scalable to large sparse networks.