Evaluating Model Performance Under Worst-case Subpopulations

TL;DR

This paper introduces a scalable method to evaluate the worst-case performance of machine learning models across subpopulations defined by core attributes, addressing distributional robustness and intersectionality.

Contribution

It develops a two-stage estimation procedure with finite-sample guarantees that assesses model robustness over complex subpopulations, considering continuous attributes and intersectionality.

Findings

Method certifies model robustness on real datasets.

Procedure provides finite-sample convergence guarantees.

Evaluation error depends on attribute dimension and out-of-sample performance.

Abstract

The performance of ML models degrades when the training population is different from that seen under operation. Towards assessing distributional robustness, we study the worst-case performance of a model over all subpopulations of a given size, defined with respect to core attributes Z. This notion of robustness can consider arbitrary (continuous) attributes Z, and automatically accounts for complex intersectionality in disadvantaged groups. We develop a scalable yet principled two-stage estimation procedure that can evaluate the robustness of state-of-the-art models. We prove that our procedure enjoys several finite-sample convergence guarantees, including dimension-free convergence. Instead of overly conservative notions based on Rademacher complexities, our evaluation error depends on the dimension of Z only through the out-of-sample error in estimating the performance conditional on Z. On real datasets, we demonstrate that our method certifies the robustness of a model and prevents deployment of unreliable models.