Anisotropic function estimation using multi-bandwidth Gaussian processes

Abstract

In nonparametric regression problems involving multiple predictors, there is typically interest in estimating an anisotropic multivariate regression surface in the important predictors while discarding the unimportant ones. Our focus is on defining a Bayesian procedure that leads to the minimax optimal rate of posterior contraction (up to a log factor) adapting to the unknown dimension and anisotropic smoothness of the true surface. We propose such an approach based on a Gaussian process prior with dimension-specific scalings, which are assigned carefully-chosen hyperpriors. We additionally show that using a homogenous Gaussian process with a single bandwidth leads to a sub-optimal rate in anisotropic cases.

Published in at http://dx.doi.org/10.1214/13-AOS1192 the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)

References in corpus (11)

• Rates of contraction of posterior distributions based on Gaussian process priors
• Adaptive Bayesian estimation using a Gaussian random field with inverse Gamma bandwidth
• Reproducing kernel Hilbert spaces of Gaussian priors
• Adaptive Bayesian multivariate density estimation with Dirichlet mixtures
• Variable Selection for Nonparametric Gaussian Process Priors: Models and Computational Strategies
• Lower bounds for posterior rates with Gaussian process priors
• Adaptive nonparametric Bayesian inference using location-scale mixture priors
• Convergence rates for Bayesian density estimation of infinite-dimensional exponential families
• Dimension adaptability of Gaussian process models with variable selection and projection
• Saddle-point integration of $C_{\infty }$ "bump" functions
• Nonparametric Bayesian model selection and averaging

Cited by in corpus (13)

• Minimax-optimal nonparametric regression in high dimensions
• Frequentist coverage and sup-norm convergence rate in Gaussian process regression
• Deep learning is adaptive to intrinsic dimensionality of model smoothness in anisotropic Besov space
• Optimality of Poisson processes intensity learning with Gaussian processes
• Comparing and weighting imperfect models using D-probabilities
• Dimension adaptability of Gaussian process models with variable selection and projection
• Optimal Bayesian estimation in random covariate design with a rescaled Gaussian process prior
• Additive Gaussian Process Regression
• Optimal estimation of variance in nonparametric regression with random design
• A Diffusion Process Perspective on Posterior Contraction Rates for Parameters
• Graph Based Gaussian Processes on Restricted Domains
• Locally Adaptive Bayes Nonparametric Regression via Nested Gaussian Processes
• Functional Gaussian Process Model for Bayesian Nonparametric Analysis