The Struggles of Feature-Based Explanations: Shapley Values vs. Minimal Sufficient Subsets
This work highlights foundational issues in explainable AI, revealing that current methods may be incomplete or misaligned, which is critical for ensuring trust and fairness in AI systems.
The paper tackles the problem of feature-based explanations for neural models by showing that even trivial models can have multiple ground-truth explanations, and that Shapley explainers and minimal sufficient subsets explainers target different types, complicating explainer development and selection.
For neural models to garner widespread public trust and ensure fairness, we must have human-intelligible explanations for their predictions. Recently, an increasing number of works focus on explaining the predictions of neural models in terms of the relevance of the input features. In this work, we show that feature-based explanations pose problems even for explaining trivial models. We show that, in certain cases, there exist at least two ground-truth feature-based explanations, and that, sometimes, neither of them is enough to provide a complete view of the decision-making process of the model. Moreover, we show that two popular classes of explainers, Shapley explainers and minimal sufficient subsets explainers, target fundamentally different types of ground-truth explanations, despite the apparently implicit assumption that explainers should look for one specific feature-based explanation. These findings bring an additional dimension to consider in both developing and choosing explainers.