Learning-Augmented Robust Algorithmic Recourse
For machine learning practitioners deploying models that may update, this work provides a method to reduce the cost of recourse when future model changes are predictable, while still limiting cost when predictions are inaccurate.
This paper introduces learning-augmented algorithmic recourse, using predictions of future model changes to reduce recourse cost while maintaining robustness. The proposed algorithm achieves a consistency-robustness trade-off, with theoretical analysis showing how prediction accuracy affects performance.
Algorithmic recourse provides individuals who receive undesirable outcomes from machine learning systems with minimum-cost improvements to achieve a desirable outcome. However, machine learning models often get updated, so the recourse may not lead to the desired outcome. The robust recourse framework chooses recourses that are less sensitive to adversarial model changes, but this comes at a higher cost. To address this, we initiate the study of learning-augmented algorithmic recourse and evaluate the extent to which a designer equipped with a prediction of the future model can reduce the cost of recourse when the prediction is accurate (consistency) while also limiting the cost even when the prediction is inaccurate (robustness). We propose a novel algorithm, study the robustness-consistency trade-off, and analyze how prediction accuracy affects performance.