Learning Models for Actionable Recourse
This addresses the need for fair and transparent decision-making in high-stakes domains like finance and law, offering a method to ensure individuals can understand and change outcomes.
The paper tackles the problem of providing actionable recourse to individuals adversely impacted by machine learning decisions, such as loan denials, by proposing a novel algorithm that guarantees recourse with high probability while maintaining accuracy.
As machine learning models are increasingly deployed in high-stakes domains such as legal and financial decision-making, there has been growing interest in post-hoc methods for generating counterfactual explanations. Such explanations provide individuals adversely impacted by predicted outcomes (e.g., an applicant denied a loan) with recourse -- i.e., a description of how they can change their features to obtain a positive outcome. We propose a novel algorithm that leverages adversarial training and PAC confidence sets to learn models that theoretically guarantee recourse to affected individuals with high probability without sacrificing accuracy. We demonstrate the efficacy of our approach via extensive experiments on real data.