SHAP-IQ: Unified Approximation of any-order Shapley Interactions
This addresses the need for efficient and reliable interaction explanations in black-box models, offering a unified solution for researchers and practitioners, though it is incremental as it builds on existing Shapley value methods.
The authors tackled the problem of approximating Shapley interactions for explainable AI by proposing SHAP-IQ, a unified sampling-based method that works for any cardinal interaction index, providing theoretical guarantees and variance estimates, and demonstrating efficiency in language, image, and synthetic models.
Predominately in explainable artificial intelligence (XAI) research, the Shapley value (SV) is applied to determine feature attributions for any black box model. Shapley interaction indices extend the SV to define any-order feature interactions. Defining a unique Shapley interaction index is an open research question and, so far, three definitions have been proposed, which differ by their choice of axioms. Moreover, each definition requires a specific approximation technique. Here, we propose SHAPley Interaction Quantification (SHAP-IQ), an efficient sampling-based approximator to compute Shapley interactions for arbitrary cardinal interaction indices (CII), i.e. interaction indices that satisfy the linearity, symmetry and dummy axiom. SHAP-IQ is based on a novel representation and, in contrast to existing methods, we provide theoretical guarantees for its approximation quality, as well as estimates for the variance of the point estimates. For the special case of SV, our approach reveals a novel representation of the SV and corresponds to Unbiased KernelSHAP with a greatly simplified calculation. We illustrate the computational efficiency and effectiveness by explaining language, image classification and high-dimensional synthetic models.