The Robustness of Estimator Composition

TL;DR

This paper investigates the robustness of composite estimators by analyzing how their breakdown points relate to those of individual estimators, providing a theoretical foundation and practical insights for robust data analysis.

Contribution

It formalizes the breakdown point for composite estimators and proves that, under mild conditions, the breakdown point of the composite is the product of individual breakdown points.

Findings

Breakdown point of composite estimators equals the product of individual breakdown points.

Applicable to various scenarios like regression and statistical testing.

Provides insights into worst-case robustness of composite estimators.

Abstract

We formalize notions of robustness for composite estimators via the notion of a breakdown point. A composite estimator successively applies two (or more) estimators: on data decomposed into disjoint parts, it applies the first estimator on each part, then the second estimator on the outputs of the first estimator. And so on, if the composition is of more than two estimators. Informally, the breakdown point is the minimum fraction of data points which if significantly modified will also significantly modify the output of the estimator, so it is typically desirable to have a large breakdown point. Our main result shows that, under mild conditions on the individual estimators, the breakdown point of the composite estimator is the product of the breakdown points of the individual estimators. We also demonstrate several scenarios, ranging from regression to statistical testing, where this analysis is easy to apply, useful in understanding worst case robustness, and sheds powerful insights onto the associated data analysis.