Improved spectral community detection in large heterogeneous networks

12 pages, 3 figures, Accepted for publication in the IEEE Transactions on Vehicular TechnologyInternational audienceIn this article, we propose and study the performance of spectral community detection for a family of "α-normalized" adjacency matrices A, of the type D −α AD −α with D the degree matrix, in heterogeneous dense graph models. We show that the previously used normaliza-tion methods based on A or D −1 AD −1 are in general suboptimal in terms of correct recovery rates and, relying on advanced random matrix methods, we prove instead the existence of an optimal value α opt of the parameter α in our generic model; we further provide an online estimation of α opt only based on the node degrees in the graph. Numerical simulations show that the proposed method outperforms state-of-the-art spectral approaches on moderately dense to dense heterogeneous graphs