Covariate-assisted Spectral Clustering in Dynamic Networks
share
In this paper, we study the community detection problem in the dynamic stochastic blockmodel and dynamic stochastic co-blockmodel with node covariates. Covariate-assisted spectral clustering methods for estimating group membership in dynamic directed and undirected graphs are developed respectively. Weak consistency property is proved and degree correction is studied as well. Simulation shows that our new method outperforms existing spectral clustering methods for studying dynamic networks.
READ FULL TEXT