A0846
Title: Efficient analysis of latent spaces in heterogeneous networks
Authors: Yinqiu He - University of Wisconsin - Madison (United States) [presenting]
Abstract: Efficient estimation of the latent structures is studied for a collection of heterogeneous networks. A latent space model is proposed with a shared latent structure along with distinct individual structures. A procedure is developed that learns the shared space from the data. Estimation is achieved by parametric efficient score equations for the latent space parameters. Oracle error rates are derived to estimate both the shared and distinct latent space parameters simultaneously. The method and theory encompass a wide range of types of edge weights under exponential family distributions.
