cs.LGJun 30, 2026

Distributionally Robust Linear Regression With Block Lewis Weights

Authors: Naren Sarayu ManojKumar Kshitij Patel

Organizations: TTIC · Institute for Foundations of Data Science, Yale University

Abstract

We present an algorithm for the group distributionally robust (GDR) least squares problem. Given mm groups, a parameter vector in Rd\mathbb{R}^d, and stacked design matrices and responses A\mathbf{A} and b\mathbf{b}, our algorithm obtains a (1+ε)(1+\varepsilon)-multiplicative optimal solution using O~(min{rank(A),m}1/3ε2/3)\widetilde{O}(\min\{\mathsf{rank}(\mathbf{A}),m\}^{1/3}\varepsilon^{-2/3}) linear-system-solves of matrices of the form ABA\mathbf{A}^{\top}\mathbf{B}\mathbf{A} for block-diagonal B\mathbf{B}. Our technical methods follow from a recent geometric construction, block Lewis weights, that relates the empirical GDR problem to a carefully chosen least squares problem and an application of accelerated proximal methods. Our algorithm improves over known interior point methods for moderate accuracy regimes and matches the state-of-the-art guarantees for the special case of \ell_{\infty} regression. We also give algorithms that smoothly interpolate between minimizing the average least squares loss and the distributionally robust loss.

Explore similar work

CardsList
  1. Distributionally-Robust Learning to Optimize

    May 7, 2026Vinit Ranjan, Jisun Park, Bartolomeo StellatoRobust OptimizationDistributional Learning