Pseudo-likelihood methods for community detection in large sparse networks
This addresses scalability and sparsity issues in network analysis, though it is incremental as it builds on existing models with new algorithmic improvements.
The authors tackled the problem of community detection in large sparse networks by proposing a fast pseudo-likelihood method for fitting stochastic block models, which performs well on very sparse networks and is illustrated with a political blogs network.
Many algorithms have been proposed for fitting network models with communities, but most of them do not scale well to large networks, and often fail on sparse networks. Here we propose a new fast pseudo-likelihood method for fitting the stochastic block model for networks, as well as a variant that allows for an arbitrary degree distribution by conditioning on degrees. We show that the algorithms perform well under a range of settings, including on very sparse networks, and illustrate on the example of a network of political blogs. We also propose spectral clustering with perturbations, a method of independent interest, which works well on sparse networks where regular spectral clustering fails, and use it to provide an initial value for pseudo-likelihood. We prove that pseudo-likelihood provides consistent estimates of the communities under a mild condition on the starting value, for the case of a block model with two communities.