English

Minimax-optimal nonparametric regression in high dimensions

Statistics Theory 2015-04-02 v3 Statistics Theory

Abstract

Minimax L2L_2 risks for high-dimensional nonparametric regression are derived under two sparsity assumptions: (1) the true regression surface is a sparse function that depends only on d=O(logn)d=O(\log n) important predictors among a list of pp predictors, with logp=o(n)\log p=o(n); (2) the true regression surface depends on O(n)O(n) predictors but is an additive function where each additive component is sparse but may contain two or more interacting predictors and may have a smoothness level different from other components. For either modeling assumption, a practicable extension of the widely used Bayesian Gaussian process regression method is shown to adaptively attain the optimal minimax rate (up to logn\log n terms) asymptotically as both n,pn,p\to\infty with logp=o(n)\log p=o(n).

Keywords

Cite

@article{arxiv.1401.7278,
  title  = {Minimax-optimal nonparametric regression in high dimensions},
  author = {Yun Yang and Surya T. Tokdar},
  journal= {arXiv preprint arXiv:1401.7278},
  year   = {2015}
}

Comments

Published at http://dx.doi.org/10.1214/14-AOS1289 in the Annals of Statistics (http://www.imstat.org/aos/) by the Institute of Mathematical Statistics (http://www.imstat.org)

R2 v1 2026-06-22T02:56:32.089Z