References & Citations
Statistics > Methodology
Title: A semiparametric extension of the stochastic block model for longitudinal networks
(Submitted on 22 Dec 2015 (v1), last revised 25 Jul 2017 (this version, v3))
Abstract: To model recurrent interaction events in continuous time, an extension of the stochastic block model is proposed where every individual belongs to a latent group and interactions between two individuals follow a conditional inhomogeneous Poisson process with intensity driven by the individuals' latent groups. The model is shown to be identifiable and its estimation is based on a semiparametric variational expectation-maximization algorithm. Two versions of the method are developed, using either a nonparametric histogram approach (with an adaptive choice of the partition size) or kernel intensity estimators. The number of latent groups can be selected by an integrated classification likelihood criterion. Finally, we demonstrate the performance of our procedure on synthetic experiments, analyse two datasets to illustrate the utility of our approach and comment on competing methods.
Submission history
From: Catherine Matias [view email][v1] Tue, 22 Dec 2015 13:22:08 GMT (1670kb,D)
[v2] Mon, 11 Jul 2016 11:35:19 GMT (1753kb,D)
[v3] Tue, 25 Jul 2017 13:12:56 GMT (1547kb,D)
Link back to: arXiv, form interface, contact.