A0408
Title: Response prediction with convergence guarantees in multiple random graphs on unknown manifolds
Authors: Aranyak Acharyya - Johns Hopkins University (United States) [presenting]
Jesus Arroyo - Texas A&M University (United States)
Michael Clayton - University of Cambridge (United Kingdom)
Marta Zlatic - University of Cambridge (United Kingdom)
Youngser Park - Johns Hopkins University (United States)
Carey Priebe - Johns Hopkins University (United States)
Abstract: The popularity of random graphs has increased in recent times owing to their applicability in analyzing network data arising from various spheres of real life, including neuroscience, biology and social studies. The model involves a collection of graphs with a shared structure on a common set of nodes, where some of the graphs are associated with responses. Assuming that each graph corresponds to a point on a one-dimensional manifold in higher dimensional ambient space, a technique that predicts the response is proposed at an unlabeled graph by exploiting the underlying manifold structure, which is unknown and hence estimated from the data. Convergence guarantees for the method are established, and its performance is demonstrated on simulated data.
