Evaluating Robustness of Counterfactual Explanations

TL;DR

This paper investigates the stability of counterfactual explanations in decision systems, highlighting their potential instability and unfairness, and proposes using plausible counterfactuals to enhance robustness and fairness.

Contribution

The paper provides a formal and empirical analysis of counterfactual explanation robustness and introduces plausible counterfactuals as a method to improve stability and fairness.

Findings

Counterfactual explanations can be highly unstable with small input changes.

Using plausible counterfactuals enhances explanation robustness.

Plausible counterfactuals improve individual fairness in explanations.

Abstract

Transparency is a fundamental requirement for decision making systems when these should be deployed in the real world. It is usually achieved by providing explanations of the system's behavior. A prominent and intuitive type of explanations are counterfactual explanations. Counterfactual explanations explain a behavior to the user by proposing actions -- as changes to the input -- that would cause a different (specified) behavior of the system. However, such explanation methods can be unstable with respect to small changes to the input -- i.e. even a small change in the input can lead to huge or arbitrary changes in the output and of the explanation. This could be problematic for counterfactual explanations, as two similar individuals might get very different explanations. Even worse, if the recommended actions differ considerably in their complexity, one would consider such unstable (counterfactual) explanations as individually unfair. In this work, we formally and empirically study the robustness of counterfactual explanations in general, as well as under different models and different kinds of perturbations. Furthermore, we propose that plausible counterfactual explanations can be used instead of closest counterfactual explanations to improve the robustness and consequently the individual fairness of counterfactual explanations.