XAudit : A Theoretical Look at Auditing with Explanations

TL;DR

This paper provides a formal theoretical framework for using explanations in machine learning model auditing, demonstrating the effectiveness of counterfactual explanations and analyzing the utility of Anchors and decision paths.

Contribution

It formalizes the role of explanations in auditing and introduces explanation-based algorithms for auditing linear classifiers and decision trees.

Findings

Counterfactual explanations significantly aid auditing.

Anchors and decision paths are helpful on average, despite worst-case limitations.

Theoretical insights into explanation utility for model auditing.

Abstract

Responsible use of machine learning requires models to be audited for undesirable properties. While a body of work has proposed using explanations for auditing, how to do so and why has remained relatively ill-understood. This work formalizes the role of explanations in auditing and investigates if and how model explanations can help audits. Specifically, we propose explanation-based algorithms for auditing linear classifiers and decision trees for feature sensitivity. Our results illustrate that Counterfactual explanations are extremely helpful for auditing. While Anchors and decision paths may not be as beneficial in the worst-case, in the average-case they do aid a lot.