mlsanityData to 14 Oct 2026

The Inadequacy of Shapley Values for Explainability

arXiv 2302.08160 ↗First posted 16 Feb 2023cs.LG cs.AIcited by 31

Not on the map. Why?

Abstract (arXiv)

This paper develops a rigorous argument for why the use of Shapley values in explainable AI (XAI) will necessarily yield provably misleading information about the relative importance of features for predictions. Concretely, this paper demonstrates that there exist classifiers, and associated predictions, for which the relative importance of features determined by the Shapley values will incorrectly assign more importance to features that are provably irrelevant for the prediction, and less importance to features that are provably relevant for the prediction. The paper also argues that, given recent complexity results, the existence of efficient algorithms for the computation of rigorous feature attribution values in the case of some restricted classes of classifiers should be deemed unlikely at best.

Key facts. Cited by 31 papers in the map (data through 14 Oct 2026). Builds on 39 of its 108 references that are in the map. First posted on arXiv 16 Feb 2023.

Builds on

39 of 108 references.

References · 39

  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. 2019Explainable Artificial Intelligence (XAI): Concepts, Taxonomies, Opportunities and Challenges toward Responsible AIcited by 1,552
  4. 2018A Survey Of Methods For Explaining Black Box Modelscited by 973
  5. 2019The many Shapley values for model explanationcited by 344
  6. 2020Problems with Shapley-value-based explanations as feature importance measurescited by 200
  7. 2018Abduction-Based Explanations for Machine Learning Modelscited by 160
  8. 2020Improving KernelSHAP: Practical Shapley Value Estimation via Linear Regressioncited by 154
  9. 2017PMLB: A Large Benchmark Suite for Machine Learning Evaluation and Comparisoncited by 140
  10. 2020Reliable Post hoc Explanations: Modeling Uncertainty in Explainabilitycited by 125
  11. 2021FastSHAP: Real-Time Shapley Value Estimationcited by 125
  12. 2019Asymmetric Shapley values: incorporating causal knowledge into model-agnostic explainabilitycited by 114
27 more
  1. 2018L-Shapley and C-Shapley: Efficient Model Interpretation for Structured Datacited by 100
  2. 2020On The Reasons Behind Decisionscited by 100
  3. 2022The Shapley Value in Machine Learningcited by 98
  4. 2020On the Tractability of SHAP Explanationscited by 92
  5. 2020Shapley explainability on the data manifoldcited by 88
  6. 2020Model Interpretability through the Lens of Computational Complexitycited by 75
  7. 2020Towards Unifying Feature Attribution and Counterfactual Explanations: Different Means to the Same Endcited by 66
  8. 2020Explaining Naive Bayes and Other Linear Classifiers with Polynomial Time and Delaycited by 57
  9. 2021On Explaining Random Forests with SATcited by 57
  10. 2021Shapley values for feature selection: The good, the bad, and the axiomscited by 54
  11. 2021On Guaranteed Optimal Robust Explanations for NLP Modelscited by 54
  12. 2022Logic-Based Explainability in Machine Learningcited by 48
  13. 2021Local Explanations via Necessity and Sufficiency: Unifying Theory and Practicecited by 45
  14. 2022On Tackling Explanation Redundancy in Decision Treescited by 44
  15. 2021On the Computational Intelligibility of Boolean Classifierscited by 40
  16. 2021SAT-Based Rigorous Explanations for Decision Listscited by 40
  17. 2021Explanations for Monotonic Classifierscited by 39
  18. 2021On Efficiently Explaining Graph-Based Classifierscited by 37
  19. 2022On Computing Probabilistic Explanations for Decision Treescited by 35
  20. 2021On the Complexity of SHAP-Score-Based Explanations: Tractability via Knowledge Compilation and Non-Approximability Resultscited by 32
  21. 2022ASTERYX : A model-Agnostic SaT-basEd appRoach for sYmbolic and score-based eXplanationscited by 26
  22. 2022Feature Necessity & Relevancy in ML Classifier Explanationscited by 23
  23. 2021Counterfactual Shapley Additive Explanationscited by 22
  24. 2021Rational Shapley Valuescited by 18
  25. 2021XOmiVAE: an interpretable deep learning model for cancer classification using high-dimensional omics datacited by 13
  26. 2021Towards Rigorous Interpretations: a Formalisation of Feature Attributioncited by 11
  27. 2021Exact Shapley Values for Local and Model-True Explanations of Decision Tree Ensemblescited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 12; 2024: 11; 2025: 3; 2026: 5.

31 papers.

Sort:

By area:

  1. 2023Explaining Predictive Uncertainty with Information Theoretic Shapley Valuescited by 36
  2. 2023Explainability is NOT a Gamecited by 23
  3. 2023Disproving XAI Myths with Formal Methods -- Initial Resultscited by 12
  4. 2023On Formal Feature Attribution and Its Approximationcited by 11
  5. 2023Axiomatic Aggregations of Abductive Explanationscited by 10
  6. 2023Anytime Approximate Formal Feature Attributioncited by 9
  7. 2024From SHAP Scores to Feature Importance Scorescited by 8
  8. 2024Auditing Local Explanations is Hardcited by 8
  9. 2024Formal Explanations for Neuro-Symbolic AIcited by 6
  10. 2023Explaining Exchange Rate Forecasts with Macroeconomic Fundamentals Using Interpretive Machine Learningcited by 5
  11. 2024X Hacking: The Threat of Misguided AutoMLcited by 5
  12. 2024Causality from Bottom to Top: A Surveycited by 5
19 more
  1. 2023A Refutation of Shapley Values for Explainabilitycited by 4
  2. 2023Refutation of Shapley Values for XAI -- Additional Evidencecited by 4
  3. 2023A Uniform Language to Explain Decision Treescited by 4
  4. 2024Logic-Based Explainability: Past, Present & Futurecited by 2
  5. 2024SHAP scores fail pervasively even when Lipschitz succeedscited by 2
  6. 2023Self-Reinforcement Attention Mechanism For Tabular Learningcited by 1
  7. 2024Error Analysis of Shapley Value-Based Model Explanations: An Informative Perspectivecited by 1
  8. 2024Stronger Baseline Models -- A Key Requirement for Aligning Machine Learning Research with Clinical Utilitycited by 1
  9. 2024Explaining the Unexplained: Revealing Hidden Correlations for Better Interpretabilitycited by 1
  10. 2025The Explanation Game -- Rekindled (Extended Version)cited by 1
  11. 2025Feature Relevancy, Necessity and Usefulness: Complexity and Algorithmscited by 1
  12. 2023What's meant by explainable model: A Scoping Reviewcited by 0
  13. 2024Unifying Interpretability and Explainability for Alzheimer's Disease Progression Predictioncited by 0
  14. 2025Feature-Guided Neighbor Selection for Non-Expert Evaluation of Model Predictionscited by 0
  15. 2026Towards Rigorous Explainability by Feature Attributioncited by 0
  16. 2026We Need Explanation Cards to Connect Explanation Algorithms to the Real Worldcited by 0
  17. 2026Beyond Shapley: Efficient Computation of Asymmetric Shapley Valuescited by 0
  18. 2026X-WAD: eXplainable Web Anomaly Detectioncited by 0
  19. 2026Solving Hard XAI Queries Based on a Compiled Dual-Rail Encodingcited by 0