B1838
Title: Nonparametric estimation for multiple heterogeneous networks
Authors: Swati Chandna - Birkbeck, University of London (United Kingdom) [presenting]
Pierre-Andre Maugis - University College London (United Kingdom)
Abstract: Nonparametric estimation is studied for the setting where multiple networks are observed on the same set of entities (nodes), with or without covariate information. Such samples may arise in the form of replicated networks assumed to be drawn from a common distribution, or in the form of longitudinal networks observed over time or space with the network generating process varying from one network to another. For example, social interaction networks between subjects over time or on different social media platforms; in connectomics where a brain network is observed for each subject along with age, gender etc. Drawing on concepts and techniques from graph theory and embedding approaches, we show how standard nonparametric methods can be employed to lead to a simple kernel estimator. Unlike existing histogram or blockmodel approximations to graphon function, our method allows estimation of node-specific as well as network-specific heterogeneity and hence offers an easy to interpret and flexible approach.
