English

$\ell_p$ Row Sampling by Lewis Weights

Data Structures and Algorithms 2014-12-02 v1 Probability

Abstract

We give a simple algorithm to efficiently sample the rows of a matrix while preserving the p-norms of its product with vectors. Given an nn-by-dd matrix A\boldsymbol{\mathit{A}}, we find with high probability and in input sparsity time an A\boldsymbol{\mathit{A}}' consisting of about dlogdd \log{d} rescaled rows of A\boldsymbol{\mathit{A}} such that Ax1\| \boldsymbol{\mathit{A}} \boldsymbol{\mathit{x}} \|_1 is close to Ax1\| \boldsymbol{\mathit{A}}' \boldsymbol{\mathit{x}} \|_1 for all vectors x\boldsymbol{\mathit{x}}. We also show similar results for all p\ell_p that give nearly optimal sample bounds in input sparsity time. Our results are based on sampling by "Lewis weights", which can be viewed as statistical leverage scores of a reweighted matrix. We also give an elementary proof of the guarantees of this sampling process for 1\ell_1.

Keywords

Cite

@article{arxiv.1412.0588,
  title  = {$\ell_p$ Row Sampling by Lewis Weights},
  author = {Michael B. Cohen and Richard Peng},
  journal= {arXiv preprint arXiv:1412.0588},
  year   = {2014}
}
R2 v1 2026-06-22T07:17:14.540Z