mlsanityData to 7 Oct 2026

Refutation of Shapley Values for XAI -- Additional Evidence

arXiv 2310.00416 ↗First posted 30 Sep 2023cs.AI cs.LGcited by 4

Not on the map. Why?

Abstract (arXiv)

Recent work demonstrated the inadequacy of Shapley values for explainable artificial intelligence (XAI). Although to disprove a theory a single counterexample suffices, a possible criticism of earlier work is that the focus was solely on Boolean classifiers. To address such possible criticism, this paper demonstrates the inadequacy of Shapley values for families of classifiers where features are not boolean, but also for families of classifiers for which multiple classes can be picked. Furthermore, the paper shows that the features changed in any minimal l0 distance adversarial examples do not include irrelevant features, thus offering further arguments regarding the inadequacy of Shapley values for XAI.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Builds on 23 of its 47 references that are in the map. First posted on arXiv 30 Sep 2023.

Builds on

23 of 47 references.

References · 23

  1. 2017A Unified Approach to Interpreting Model Predictionscited by 6,488
  2. 2016Adversarial examples in the physical worldcited by 1,762
  3. 2017Explanation in Artificial Intelligence: Insights from the Social Sciencescited by 1,201
  4. 2019The many Shapley values for model explanationcited by 341
  5. 2019Feature relevance quantification in explainable AI: A causal problemcited by 164
  6. 2018Abduction-Based Explanations for Machine Learning Modelscited by 159
  7. 2017PMLB: A Large Benchmark Suite for Machine Learning Evaluation and Comparisoncited by 139
  8. 2018A Symbolic Approach to Explaining Bayesian Network Classifierscited by 130
  9. 2020Reliable Post hoc Explanations: Modeling Uncertainty in Explainabilitycited by 125
  10. 2018L-Shapley and C-Shapley: Efficient Model Interpretation for Structured Datacited by 100
  11. 2020On the Tractability of SHAP Explanationscited by 92
  12. 2022Logic-Based Explainability in Machine Learningcited by 47
11 more
  1. 2022On Tackling Explanation Redundancy in Decision Treescited by 44
  2. 2020Minimal Adversarial Examples for Deep Learning on 3D Point Cloudscited by 41
  3. 2021On Efficiently Explaining Graph-Based Classifierscited by 37
  4. 2017Adversarial Example Defenses: Ensembles of Weak Defenses are not Strongcited by 32
  5. 2021On the Complexity of SHAP-Score-Based Explanations: Tractability via Knowledge Compilation and Non-Approximability Resultscited by 32
  6. 2023The Inadequacy of Shapley Values for Explainabilitycited by 30
  7. 2023Explainability is NOT a Gamecited by 23
  8. 2022Feature Necessity & Relevancy in ML Classifier Explanationscited by 22
  9. 2023From Robustness to Explainability and Back Againcited by 14
  10. 2023Disproving XAI Myths with Formal Methods -- Initial Resultscited by 12
  11. 2023A Refutation of Shapley Values for Explainabilitycited by 4

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 2; 2025: 0; 2026: 2.

4 papers.

  1. 2024From SHAP Scores to Feature Importance Scorescited by 8
  2. 2024Logic-Based Explainability: Past, Present & Futurecited by 2
  3. 2026Binary Spiking Neural Networks as Causal Modelscited by 1
  4. 2026Towards Rigorous Explainability by Feature Attributioncited by 0