Non-Asymptotic Bounds for the $\ell_{\infty}$ Estimator in Linear Regression with Uniform Noise

TL;DR

This paper derives non-asymptotic bounds for the $ ext{l}_ ext{infinity}$ estimator in linear regression with uniform noise, demonstrating near-minimax optimality in some designs and advantages over LASSO in high dimensions.

Contribution

It provides the first finite-sample error bounds for the Chebyshev estimator under uniform noise and compares its performance to LASSO in high-dimensional settings.

Findings

Error rate bounded by C_p/n with high probability

Existence of designs where the estimator is nearly minimax optimal

Chebyshev's LASSO outperforms regular LASSO under certain conditions

Abstract

The Chebyshev or $\ell_{\infty}$ estimator is an unconventional alternative to the ordinary least squares in solving linear regressions. It is defined as the minimizer of the $\ell_{\infty}$ objective function \begin{align*} \hat{\boldsymbol{\beta}} := \arg\min_{\boldsymbol{\beta}} \|\boldsymbol{Y} - \mathbf{X}\boldsymbol{\beta}\|_{\infty}. \end{align*} The asymptotic distribution of the Chebyshev estimator under fixed number of covariates was recently studied (Knight, 2020), yet finite sample guarantees and generalizations to high-dimensional settings remain open. In this paper, we develop non-asymptotic upper bounds on the estimation error $\|\hat{\boldsymbol{\beta}}-\boldsymbol{\beta}^*\|_2$ for a Chebyshev estimator $\hat{\boldsymbol{\beta}}$, in a regression setting with uniformly distributed noise $\varepsilon_i\sim U([-a,a])$ where $a$ is either known or unknown. With relatively mild assumptions on the (random) design matrix $\mathbf{X}$, we can bound the error rate by $\frac{C_p}{n}$ with high probability, for some constant $C_p$ depending on the dimension $p$ and the law of the design. Furthermore, we illustrate that there exist designs for which the Chebyshev estimator is (nearly) minimax optimal. On the other hand we also argue that there exist designs for which this estimator behaves sub-optimally in terms of the constant $C_p$'s dependence on $p$. In addition we show that "Chebyshev's LASSO" has advantages over the regular LASSO in high dimensional situations, provided that the noise is uniform. Specifically, we argue that it achieves a much faster rate of estimation under certain assumptions on the growth rate of the sparsity level and the ambient dimension with respect to the sample size.