English

Nonparametric Online Regression while Learning the Metric

Machine Learning 2017-10-24 v2

Abstract

We study algorithms for online nonparametric regression that learn the directions along which the regression function is smoother. Our algorithm learns the Mahalanobis metric based on the gradient outer product matrix G\boldsymbol{G} of the regression function (automatically adapting to the effective rank of this matrix), while simultaneously bounding the regret ---on the same data sequence--- in terms of the spectrum of G\boldsymbol{G}. As a preliminary step in our analysis, we extend a nonparametric online learning algorithm by Hazan and Megiddo enabling it to compete against functions whose Lipschitzness is measured with respect to an arbitrary Mahalanobis metric.

Keywords

Cite

@article{arxiv.1705.07853,
  title  = {Nonparametric Online Regression while Learning the Metric},
  author = {Ilja Kuzborskij and Nicolò Cesa-Bianchi},
  journal= {arXiv preprint arXiv:1705.07853},
  year   = {2017}
}

Comments

To appear in NIPS 2017

R2 v1 2026-06-22T19:55:06.817Z