Stochastic expansions using continuous dictionaries: Lévy adaptive regression kernels

DOI10.1214/11-AOS889MaRDI QIDQ98918zbMATHOpenAlexFDO

Authors Robert L. Wolpert, Merlise Clyde, Chong Tu, Robert L. Wolpert, Merlise A. Clyde, Chong Tu

Publication date 1 August 2011

Published in The Annals of Statistics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1112.3149

wavelets splines kernel regression nonparametric regression Bayes Besov LARK relevance vector machine reversible jump Markov chain Monte Carlo support vector machine Lévy random field

Mathematics Subject Classification ID

Bayesian inference (62F15) Density estimation (62G07) Nonparametric regression and quantile regression (62G08) Random fields; image analysis (62M40) Monte Carlo methods (65C05) Nontrigonometric harmonic analysis involving wavelets and other special systems (42C40)

Abstract: This article describes a new class of prior distributions for nonparametric function estimation. The unknown function is modeled as a limit of weighted sums of kernels or generator functions indexed by continuous parameters that control local and global features such as their translation, dilation, modulation and shape. L'{e}vy random fields and their stochastic integrals are employed to induce prior distributions for the unknown functions or, equivalently, for the number of kernels and for the parameters governing their features. Scaling, shape, and other features of the generating functions are location-specific to allow quite different function properties in different parts of the space, as with wavelet bases and other methods employing overcomplete dictionaries. We provide conditions under which the stochastic expansions converge in specified Besov or Sobolev norms. Under a Gaussian error model, this may be viewed as a sparse regression problem, with regularization induced via the L'{e}vy random field prior distribution. Posterior inference for the unknown functions is based on a reversible jump Markov chain Monte Carlo algorithm. We compare the L'{e}vy Adaptive Regression Kernel (LARK) method to wavelet-based methods using some of the standard test functions, and illustrate its flexibility and adaptability in nonstationary applications.

Recommendations

Cites work

Cited in

(8)

Describes a project that uses

Uses Software

This page was built for publication: Stochastic expansions using continuous dictionaries: Lévy adaptive regression kernels

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q98918)