mlsanityData to 7 Oct 2026

Shapley values for feature selection: The good, the bad, and the axioms

arXiv 2102.10936 ↗First posted 22 Feb 2021cs.LG cs.AI stat.MLcited by 54

Not on the map. Why?

Abstract (arXiv)

The Shapley value has become popular in the Explainable AI (XAI) literature, thanks, to a large extent, to a solid theoretical foundation, including four "favourable and fair" axioms for attribution in transferable utility games. The Shapley value is provably the only solution concept satisfying these axioms. In this paper, we introduce the Shapley value and draw attention to its recent uses as a feature selection tool. We call into question this use of the Shapley value, using simple, abstract "toy" counterexamples to illustrate that the axioms may work against the goals of feature selection. From this, we develop a number of insights that are then investigated in concrete simulation settings, with a variety of Shapley value formulations, including SHapley Additive exPlanations (SHAP) and Shapley Additive Global importancE (SAGE).

Key facts. Cited by 54 papers in the map (data through 7 Oct 2026). Builds on 15 of its 54 references that are in the map. First posted on arXiv 22 Feb 2021.

Builds on

15 of 54 references.

References · 15

  1. 2017A Unified Approach to Interpreting Model Predictionscited by 6,488
  2. 2019The many Shapley values for model explanationcited by 341
  3. 2018Consistent Individualized Feature Attribution for Tree Ensemblescited by 279
  4. 2020Explaining by Removing: A Unified Framework for Model Explanationcited by 218
  5. 2019Explaining individual predictions when features are dependent: More accurate approximations to Shapley valuescited by 205
  6. 2020Problems with Shapley-value-based explanations as feature importance measurescited by 198
  7. 2019Asymmetric Shapley values: incorporating causal knowledge into model-agnostic explainabilitycited by 114
  8. 2019Explainable AI for Trees: From Local Explanations to Global Understandingcited by 64
  9. 2020Efficient computation and analysis of distributional Shapley valuescited by 38
  10. 2020Explaining predictive models with mixed features using Shapley values and conditional inference treescited by 18
  11. 2020Predictive and Causal Implications of using Shapley Value for Model Interpretationcited by 16
  12. 2019The Explanation Game: Explaining Machine Learning Models Using Shapley Valuescited by 8
3 more
  1. 2018A Feature Selection Method Based on Shapley Value to False Alarm Reduction in ICUs, A Genetic-Algorithm Approachcited by 7
  2. 2020Explaining the data or explaining a model? Shapley values that uncover non-linear dependenciescited by 4
  3. 2020Interpretable feature subset selection: A Shapley value based approachcited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 3; 2022: 8; 2023: 9; 2024: 13; 2025: 12; 2026: 9.

54 papers.

Sort:

By area:

  1. 2024AttnLRP: Attention-Aware Layer-Wise Relevance Propagation for Transformerscited by 116
  2. 2022The Shapley Value in Machine Learningcited by 97
  3. 2024Explainable AI for Safe and Trustworthy Autonomous Driving: A Systematic Reviewcited by 52
  4. 2023The Inadequacy of Shapley Values for Explainabilitycited by 30
  5. 2021To trust or not to trust an explanation: using LEAF to evaluate local linear XAI methodscited by 26
  6. 2024Provably Accurate Shapley Value Estimation via Leverage Score Samplingcited by 23
  7. 2021Using Shapley Values and Variational Autoencoders to Explain Predictive Models with Dependent Mixed Featurescited by 19
  8. 2023A Comparative Study of Methods for Estimating Conditional Shapley Values and When to Use Themcited by 15
  9. 2024Analyzing the Role of Semantic Representations in the Era of Large Language Modelscited by 12
  10. 2024SyntaxShap: Syntax-aware Explainability Method for Text Generationcited by 11
  11. 2022New Definitions and Evaluations for Saliency Methods: Staying Intrinsic, Complete and Soundcited by 10
  12. 2023Prompt Valuation Based on Shapley Valuescited by 10
38 more
  1. 2024ROSE: A Reward-Oriented Data Selection Framework for LLM Task-Specific Instruction Tuningcited by 10
  2. 2025INTER: Mitigating Hallucination in Large Vision-Language Models by Interaction Guidance Samplingcited by 10
  3. 2022Differentially Private Shapley Values for Data Evaluationcited by 8
  4. 2025SHAP Meets Tensor Networks: Provably Tractable Explanations with Parallelismcited by 8
  5. 2022SHAP-XRT: The Shapley Value Meets Conditional Independence Testingcited by 6
  6. 2025Revisiting LRP: Positional Attribution as the Missing Ingredient for Transformer Explainabilitycited by 6
  7. 2024Explainability of Machine Learning Models under Missing Datacited by 5
  8. 2022A Scalable, Interpretable, Verifiable & Differentiable Logic Gate Convolutional Neural Network Architecture From Truth Tablescited by 4
  9. 2022Predicting Dynamic Stability from Static Features in Power Grid Models using Machine Learningcited by 4
  10. 2023A Refutation of Shapley Values for Explainabilitycited by 4
  11. 2022Local Interpretable Model Agnostic Shap Explanations for machine learning modelscited by 3
  12. 2021Inferring feature importance with uncertainties in high-dimensional datacited by 2
  13. 2023Shapley Based Residual Decomposition for Instance Analysiscited by 2
  14. 2023Beyond Single-Feature Importance with ICECREAMcited by 2
  15. 2024Explainable AI for survival analysis: a median-SHAP approachcited by 2
  16. 2024A Comprehensive Study of Shapley Value in Data Analyticscited by 2
  17. 2025From Pixels to Perception: Interpretable Predictions via Instance-wise Grouped Feature Selectioncited by 2
  18. 2025Enhancing Interpretability and Effectiveness in Recommendation with Numerical Features via Learning to Contrast the Counterfactual samplescited by 2
  19. 2025Quantifying Genuine Awareness in Hallucination Prediction Beyond Question-Side Shortcutscited by 2
  20. 2023Investigating Poor Performance Regions of Black Boxes: LIME-based Exploration in Sepsis Detectioncited by 1
  21. 2024REFRESH: Responsible and Efficient Feature Reselection Guided by SHAP Valuescited by 1
  22. 2024Energy-Based Model for Accurate Estimation of Shapley Values in Feature Attributioncited by 1
  23. 2024Shap-Select: Lightweight Feature Selection Using SHAP Values and Regressioncited by 1
  24. 2025Explaining Deep Learning-based Anomaly Detection in Energy Consumption Data by Focusing on Contextually Relevant Datacited by 1
  25. 2025Feature Relevancy, Necessity and Usefulness: Complexity and Algorithmscited by 1
  26. 2025Pattern-Based Graph Classification: Comparison of Quality Measures and Importance of Preprocessingcited by 1
  27. 2026MinShap: A Shapley-Based Framework for Feature Redundancycited by 1
  28. 2022Data-Driven Evaluation of Training Action Space for Reinforcement Learningcited by 0
  29. 2023Explainable Machine Learning for Categorical and Mixed Data with Lossless Visualizationcited by 0
  30. 2023On the Connection between Game-Theoretic Feature Attributions and Counterfactual Explanationscited by 0
  31. 2024Identifying Student Profiles Within Online Judge Systems Using Explainable Artificial Intelligencecited by 0
  32. 2025Are We Merely Justifying Results ex Post Facto? Quantifying Explanatory Inversion in Post-Hoc Model Explanationscited by 0
  33. 2025Rigorous Feature Importance Scores based on Shapley Value and Banzhaf Indexcited by 0
  34. 2025Beyond Additivity: Sparse Isotonic Shapley Regression toward Nonlinear Explainabilitycited by 0
  35. 2026Statistical Inference for Explainable Boosting Machinescited by 0
  36. 2026Evolutionary Router Feature Generation for Zero-Shot Graph Anomaly Detection with Mixture-of-Expertscited by 0
  37. 2026Verified SHAP: Provable Bounds for Exact Shapley Values of Neural Networkscited by 0
  38. 2026Beyond Shapley: Efficient Computation of Asymmetric Shapley Valuescited by 0

50 most cited of 54 All 54 (JSON)