Title: Function estimation on a large graph using Bayesian Laplacian regularization
Authors: Alisa Kirichenko - CWI (Centrum Wiskunde & Informatica) (Netherlands) [presenting]
Harry Zanten - University of Amsterdam (Netherlands)
Abstract: In recent years there has been substantial interest in high-dimensional estimation and prediction problems on large graphs. These can in many cases be viewed as high-dimensional or nonparametric regression or classification problems in which the goal is to learn a ``smooth'' function on a given graph. We present a mathematical framework that allows us to study the performance of nonparametric function estimation methods on large graphs and we derive minimax convergence rates within the framework. We consider simple undirected graphs that satisfy an assumption on their ``asymptotic geometry'', formulated in terms of the graph Laplacian. We also introduce a Sobolev-type smoothness condition on the target function using the graph Laplacian to quantify smoothness. Then we develop Bayesian procedures for problems at hand and we show how asymptotically optimal Bayesian regularization can be achieved under these conditions. The priors we study are randomly scaled Gaussians with precision operators involving the Laplacian of the graph.