English

$\ell_{\infty}$ Vector Contraction for Rademacher Complexity

Machine Learning 2019-11-18 v1 Statistics Theory Machine Learning Statistics Theory

Abstract

We show that the Rademacher complexity of any RK\mathbb{R}^{K}-valued function class composed with an \ell_{\infty}-Lipschitz function is bounded by the maximum Rademacher complexity of the restriction of the function class along each coordinate, times a factor of O~(K)\tilde{O}(\sqrt{K}).

Cite

@article{arxiv.1911.06468,
  title  = {$\ell_{\infty}$ Vector Contraction for Rademacher Complexity},
  author = {Dylan J. Foster and Alexander Rakhlin},
  journal= {arXiv preprint arXiv:1911.06468},
  year   = {2019}
}

Comments

Technical note

R2 v1 2026-06-23T12:16:46.237Z