mlsanityData to 14 Oct 2026

Feature Responsiveness Scores: Model-Agnostic Explanations for Recourse

arXiv 2410.22598 ↗First posted 29 Oct 2024stat.ML cs.LGcited by 8

Not on the map. Why?

Abstract (arXiv)

Consumer protection rules require companies that deploy models to automate decisions in high-stakes settings to explain predictions to decision subjects. These rules are motivated, in part, by the belief that explanations can promote recourse by revealing information that decision subjects can use to contest or overturn their predictions. In practice, companies provide individuals with a list of principal reasons based on feature importance derived from methods like SHAP and LIME. In this work, we show how common practices can fail to provide recourse and propose to highlight features based on their responsiveness -- the probability that a decision subject can attain a target prediction through an arbitrary intervention on the feature. We develop efficient methods to compute responsiveness scores for any model and actionability constraints. We show that standard practices in lending can undermine decision subjects by highlighting unresponsive features and explaining predictions that are fixed.

Key facts. Cited by 8 papers in the map (data through 14 Oct 2026). Builds on 45 of its 91 references that are in the map. First posted on arXiv 29 Oct 2024.

Builds on

45 of 91 references.

References · 45

  1. 2017A Unified Approach to Interpreting Model Predictionscited by 6,555
  2. 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,070
  3. 2016Equality of Opportunity in Supervised Learningcited by 2,007
  4. 2020Concept Bottleneck Modelscited by 1,032
  5. 2016Distribution-Free Predictive Inference For Regressioncited by 668
  6. 2019Explaining Machine Learning Classifiers through Diverse Counterfactual Explanationscited by 614
  7. 2019Fooling LIME and SHAP: Adversarial Attacks on Post hoc Explanation Methodscited by 443
  8. 2018Actionable Recourse in Linear Classificationcited by 328
  9. 2015Strategic Classificationcited by 257
  10. 2020Algorithmic Recourse: from Counterfactual Explanations to Interventionscited by 250
  11. 2019Model-Agnostic Counterfactual Explanations for Consequential Decisionscited by 204
  12. 2019Mitigating Bias in Algorithmic Hiring: Evaluating Claims and Practicescited by 174
33 more
  1. 2019Predictive Multiplicity in Classificationcited by 166
  2. 2022The Disagreement Problem in Explainable Machine Learning: A Practitioner's Perspectivecited by 159
  3. 2018Aequitas: A Bias and Fairness Audit Toolkitcited by 147
  4. 2020Debugging Tests for Model Explanationscited by 147
  5. 2020Algorithmic recourse under imperfect causal knowledge: a probabilistic approachcited by 140
  6. 2021FastSHAP: Real-Time Shapley Value Estimationcited by 125
  7. 2020A survey of algorithmic recourse: definitions, formulations, solutions, and prospectscited by 118
  8. 2019The Hidden Assumptions Behind Counterfactual Explanations and Principal Reasonscited by 115
  9. 2021Counterfactual Explanations Can Be Manipulatedcited by 113
  10. 2021Towards Robust and Reliable Algorithmic Recoursecited by 102
  11. 2021Do Feature Attribution Methods Correctly Attribute Features?cited by 98
  12. 2019Fairwashing: the risk of rationalizationcited by 81
  13. 2022Impossibility Theorems for Feature Attributioncited by 78
  14. 2020On the Fairness of Causal Algorithmic Recoursecited by 70
  15. 2019DLIME: A Deterministic Local Interpretable Model-Agnostic Explanations Approach for Computer-Aided Diagnosis Systemscited by 59
  16. 2021On the Adversarial Robustness of Causal Algorithmic Recoursecited by 56
  17. 2021Explaining Black-Box Algorithms Using Probabilistic Contrastive Counterfactualscited by 55
  18. 2021S-LIME: Stabilized-LIME for Model Explanationcited by 52
  19. 2023SHAP-IQ: Unified Approximation of any-order Shapley Interactionscited by 51
  20. 2022Predictive Multiplicity in Probabilistic Classificationcited by 44
  21. 2016Auditing Black-box Models for Indirect Influencecited by 41
  22. 2022Probabilistically Robust Recourse: Navigating the Trade-offs between Costs and Robustness in Algorithmic Recoursecited by 40
  23. 2023Algorithmic Collective Action in Machine Learningcited by 31
  24. 2022Attribution-based Explanations that Provide Recourse Cannot be Robustcited by 20
  25. 2019bLIMEy: Surrogate Prediction Explanations Beyond LIMEcited by 17
  26. 2023Distributionally Robust Recourse Actioncited by 12
  27. 2019Disentangling Influence: Using Disentangled Representations to Audit Model Predictionscited by 11
  28. 2022Personalized Algorithmic Recourse with Preference Elicitationcited by 9
  29. 2024Exploiting Preference Elicitation in Interactive and User-centered Algorithmic Recourse: An Initial Explorationcited by 8
  30. 2023Manipulation Risks in Explainable AI: The Implications of the Disagreement Problemcited by 7
  31. 2023Confident Feature Rankingcited by 6
  32. 2023Participatory Personalization in Classificationcited by 5
  33. 2023Prediction without Preclusion: Recourse Verification with Reachable Setscited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 4; 2026: 4.

8 papers.

  1. 2025Reinforcement Learning for Durable Algorithmic Recoursecited by 3
  2. 2026Fair Recourse for All: Ensuring Individual and Group Fairness in Counterfactual Explanationscited by 2
  3. 2025Regretful Decisions under Label Noisecited by 1
  4. 2025Explanations are a Means to an End: Decision Theoretic Explanation Evaluationcited by 1
  5. 2026From Universal to Individualized Actionability: Revisiting Personalization in Algorithmic Recoursecited by 1
  6. 2026The Role of Causality in Algorithmic Recoursecited by 1
  7. 2025Responsiveness Verification: Will Predictions Change? How Much? How Often?cited by 0
  8. 2026Algorithmic Recourse Under Competitioncited by 0