On the Connection between Game-Theoretic Feature Attributions and Counterfactual Explanations

Abstract

Explainable Artificial Intelligence (XAI) has received widespread interest in recent years, and two of the most popular types of explanations are feature attributions, and counterfactual explanations. These classes of approaches have been largely studied independently and the few attempts at reconciling them have been primarily empirical. This work establishes a clear theoretical connection between game-theoretic feature attributions, focusing on but not limited to SHAP, and counterfactuals explanations. After motivating operative changes to Shapley values based feature attributions and counterfactual explanations, we prove that, under conditions, they are in fact equivalent. We then extend the equivalency result to game-theoretic solution concepts beyond Shapley values. Moreover, through the analysis of the conditions of such equivalence, we shed light on the limitations of naively using counterfactual explanations to provide feature importances. Experiments on three datasets quantitatively show the difference in explanations at every stage of the connection between the two approaches and corroborate the theoretical findings.

arXiv commentAccepted at AIES 2023

References in corpus (24)

• XGBoost: A Scalable Tree Boosting System
• Artificial Intelligence: the global landscape of ethics guidelines
• Axiomatic Attribution for Deep Networks
• A Survey of Deep Learning Techniques for Autonomous Driving
• Explaining Machine Learning Classifiers through Diverse Counterfactual Explanations
• The role of explainability in creating trustworthy artificial intelligence for health care: a comprehensive survey of the terminology, design choices, and evaluation strategies
• Actionable Recourse in Linear Classification
• FACE: Feasible and Actionable Counterfactual Explanations
• The Hidden Assumptions Behind Counterfactual Explanations and Principal Reasons
• CERTIFAI: Counterfactual Explanations for Robustness, Transparency, Interpretability, and Fairness of Artificial Intelligence models
• Feature relevance quantification in explainable AI: A causal problem
• Towards Unifying Feature Attribution and Counterfactual Explanations: Different Means to the Same End
• Algorithmic recourse under imperfect causal knowledge: a probabilistic approach
• Counterfactual Shapley Additive Explanations
• Towards Robust and Reliable Algorithmic Recourse
• On Counterfactual Explanations under Predictive Multiplicity
• Feature Removal Is a Unifying Principle for Model Explanation Methods
• Probabilistically Robust Recourse: Navigating the Trade-offs between Costs and Robustness in Algorithmic Recourse
• Counterfactual Explanations for Arbitrary Regression Models
• Combining Counterfactuals With Shapley Values To Explain Image Models
• Feature Importance for Time Series Data: Improving KernelSHAP
• Causal Discovery and Knowledge Injection for Contestable Neural Networks (with Appendices)
• Principled Diverse Counterfactuals in Multilinear Models
• FASTER-CE: Fast, Sparse, Transparent, and Robust Counterfactual Explanations