Actionable Inverse Classification with Action Fairness Guarantees
Keywords: Algorithm Fairness, Explainability
Abstract: Machine learning (ML) classifiers are increasingly used in critical decision-making domains such as finance, healthcare, and the judiciary. However, their interpretability and fairness remain significant challenges, often leaving users without clear guidance on how to improve unfavourable outcomes. This paper introduces an actionable ML framework that provides minimal, explainable modifications to input data to change classification results. We also propose a novel concept of "action fairness," which ensures that users from different subgroups incur similar costs when altering their classification outcomes. Our approach identifies the nearest decision boundary point to a given query, allowing for the determination of minimal cost actions. We demonstrate the effectiveness of this method using real-world credit assessment data, showing that our solution not only improves the fairness of classifier outcomes but also enhances their usability and interpretability.
Primary Area: alignment, fairness, safety, privacy, and societal considerations
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 7946
Loading