On the robustness to adversarial corruption and to heavy-tailed data of the Stahel-Donoho median of means

TL;DR

This paper introduces robust, subgaussian estimators for mean vectors and covariance matrices based on median of means versions of Stahel-Donoho outlyingness and MAD, effective under adversarial contamination and heavy-tailed data.

Contribution

It provides the first non-asymptotic bounds for the Stahel-Donoho median, extending robustness analysis to cases with minimal moment assumptions and non-existent means.

Findings

Achieves subgaussian rates under heavy tails and adversarial contamination.

Provides non-asymptotic bounds for Stahel-Donoho median estimators.

Constructs covariance estimators under minimal moment assumptions.

Abstract

We consider median of means (MOM) versions of the Stahel-Donoho outlyingness (SDO) [stahel 1981, donoho 1982] and of Median Absolute Deviation (MAD) functions to construct subgaussian estimators of a mean vector under adversarial contamination and heavy-tailed data. We develop a single analysis of the MOM version of the SDO which covers all cases ranging from the Gaussian case to the L2 case. It is based on isomorphic and almost isometric properties of the MOM versions of SDO and MAD. This analysis also covers cases where the mean does not even exist but a location parameter does; in those cases we still recover the same subgaussian rates and the same price for adversarial contamination even though there is not even a first moment. These properties are achieved by the classical SDO median and are therefore the first non-asymptotic statistical bounds on the Stahel-Donoho median complementing the $\sqrt{n}$-consistency [maronna 1995] and asymptotic normality [Zuo, Cui, He, 2004] of the Stahel-Donoho estimators. We also show that the MOM version of MAD can be used to construct an estimator of the covariance matrix under only a L2-moment assumption or of a scale parameter if a second moment does not exist.