Abstract:
|
Community detection has emerged in recent years as one of the fundamental problems of network analysis. Informally, one seeks to partition the network into cohesive groups of nodes, or communities, that reveal its large-scale connective structure. In this talk, we explore community detection in the setting of bipartite networks. We consider extensions of the so-called Stochastic Block Model (SBM) to the bipartite setting and show how these models can be efficiently fit using a combination of spectral and likelihood-based approaches. In particular, we show how simple pseudo-likelihood updates can be used to boost the performance of spectral clustering and achieve optimal rates of misclassification. We also present some new results on the nature of these optimal rates in the bipartite setting, sharpening existing estimates.
|