Title: Clustering of a directed graph: Bipartite clustering or not
Authors: Christine Keribin - INRIA - Paris-Saclay University (France) [presenting]
Abstract: The Stochastic Block Model (SBM) is commonly used not only for the clustering of undirected graphs but also for the clustering of directed graphs too. This choice should be discussed because this model makes no difference between the clusters of source and target nodes. That is why the use of the Latent Block Model (LBM), building two different clusterings for the source and target nodes, could be interesting for directed graph clustering. We will analyze and discuss (structure, inference, model selection) through simulated data the comparison between SBM and LBM for directed graph clustering, and propose a methodology which will be applied on real data sets.