A0438
Title: Spectral analysis on networks with covariates
Authors: Wanjie Wang - National University of Singapore (Singapore) [presenting]
Abstract: Social network data record the connections between objects. In past decades, the study of social network data has been an important topic. Together with the observed connections, the profile of the object itself is usually also recorded. The data on the node-level profile covariates is called. Both the covariates and the network reflect the underlying data structure. The results on how the covariates will help to solve problems on the networks, such as community detection, and how the network helps with problems about the covariates, such as feature selection, are introduced. Spectral methods, which are computationally efficient in such problems, are mainly considered. Hence, the theoretical results on the control of the eigenvectors/eigenvalues are introduced when both the networks and covariates are combined. Such results are useful for further studies.
