Explainable AI methods have been proposed to help interpret complex models, e.g., by assigning importance scores to model features or perturbing the features in a way that changes the prediction. These methods apply to one ...