No abstract

In dynamic network analysis, there has been a surge of interest in community detection and evolution. However, existing methods for dynamic community detection do not consider the dependency among edges, which could lead to information loss in detecting community structures. In this paper, we investigate the problem of identifying a change-point with abrupt changes in community structure of a network. We propose an approximate likelihood approach for the change-point estimator as well as node membership identification through integrating both marginal information and dependency of network connectivities. An EM-type algorithm is proposed for maximizing the approximate likelihood jointly over change-point and community membership evolution. In theory, we establish estimation consistency under the regularity condition and show that the proposed estimators achieve a higher convergence rate compared with their marginal likelihood counterpart without incorporating dependency among edges.The validity of the proposed method is supported via application on the ADHD-200 dataset for detecting brain functional community change over time.