Boundary-Aware Uncertainty for Feature Attribution Explainers
This work addresses the need for trustworthy explanations in high-stakes applications, offering a flexible method to improve understanding of feature attribution explainers.
The authors tackled the problem of unreliable local explanations for black-box classifiers by proposing the GPEC framework, which quantifies explanation uncertainty by combining decision boundary-aware and approximation uncertainties, and demonstrated its effectiveness on multiple datasets.
Post-hoc explanation methods have become a critical tool for understanding black-box classifiers in high-stakes applications. However, high-performing classifiers are often highly nonlinear and can exhibit complex behavior around the decision boundary, leading to brittle or misleading local explanations. Therefore there is an impending need to quantify the uncertainty of such explanation methods in order to understand when explanations are trustworthy. In this work we propose the Gaussian Process Explanation UnCertainty (GPEC) framework, which generates a unified uncertainty estimate combining decision boundary-aware uncertainty with explanation function approximation uncertainty. We introduce a novel geodesic-based kernel, which captures the complexity of the target black-box decision boundary. We show theoretically that the proposed kernel similarity increases with decision boundary complexity. The proposed framework is highly flexible; it can be used with any black-box classifier and feature attribution method. Empirical results on multiple tabular and image datasets show that the GPEC uncertainty estimate improves understanding of explanations as compared to existing methods.