STA / BST 290 Seminar Series
Wednesday, January 21, 4:10pm, MSB 1147 (Colloquium Room)
Speaker: Rachel Wang (University of California, Berkeley)
Title:Problems in network modeling: inferring edges and community detection
Abstract:Networks pervade many disciplines of science as a way of analyzing complex systems with interacting components. The problem of network modeling is often two-fold. First, the relationships between pairs of nodes, if not directly observed, have to be estimated from data. Based on the estimated (or given) network topology, various statistical and computational tools can then be applied to extract interesting patterns such as the presence of communities. In this talk I will present two studies related to both parts of the problem. I will first discuss a study in the context of gene regulatory networks, where the goal is to infer gene interactions using expression data with large and heterogeneous samples. We propose two gene coexpression statistics based on counting local patterns of expression ranks to take into account the potentially changing nature of gene relationships. Moving onto general networks, I will then discuss model selection for the stochastic block model (SBM), which is a popular tool for community detection. We consider an approach based on the log likelihood ratio statistic and analyze its asymptotic properties under model misspecification. The results enable us to derive the correct order of the penalty term for model complexity and arrive at a likelihood-based model selection criterion that is asymptotically consistent and valid also in the semi-sparse regime.