The Explanation Game -- Rekindled (Extended Version)
This addresses the reliability of explainable AI for users who depend on accurate interpretations, though it is incremental as it builds on existing Shapley value methods.
The paper tackles the problem of critical flaws in current Shapley value-based explanations in XAI, which can mislead human decision-makers, by proposing a novel definition of SHAP scores that overcomes these flaws and providing an efficient estimation method, with preliminary experimental results confirming the claims.
Recent work demonstrated the existence of critical flaws in the current use of Shapley values in explainable AI (XAI), i.e. the so-called SHAP scores. These flaws are significant in that the scores provided to a human decision-maker can be misleading. Although these negative results might appear to indicate that Shapley values ought not be used in XAI, this paper argues otherwise. Concretely, this paper proposes a novel definition of SHAP scores that overcomes existing flaws. Furthermore, the paper outlines a practically efficient solution for the rigorous estimation of the novel SHAP scores. Preliminary experimental results confirm our claims, and further underscore the flaws of the current SHAP scores.