Pixel-level Certified Explanations via Randomized Smoothing

TL;DR

This paper introduces a novel certification framework using randomized smoothing to provide pixel-level robustness guarantees for any black-box attribution method, enhancing trustworthiness and interpretability of explanations in deep learning.

Contribution

It presents the first method to certify pixel-level robustness for attribution maps, reformulating the problem as segmentation and proposing new evaluation metrics.

Findings

Certified attributions are robust against $\

The framework applies to 12 attribution methods across 5 ImageNet models.

The approach improves interpretability and faithfulness of explanations.

Abstract

Post-hoc attribution methods aim to explain deep learning predictions by highlighting influential input pixels. However, these explanations are highly non-robust: small, imperceptible input perturbations can drastically alter the attribution map while maintaining the same prediction. This vulnerability undermines their trustworthiness and calls for rigorous robustness guarantees of pixel-level attribution scores. We introduce the first certification framework that guarantees pixel-level robustness for any black-box attribution method using randomized smoothing. By sparsifying and smoothing attribution maps, we reformulate the task as a segmentation problem and certify each pixel's importance against $\ell_2$-bounded perturbations. We further propose three evaluation metrics to assess certified robustness, localization, and faithfulness. An extensive evaluation of 12 attribution methods across 5 ImageNet models shows that our certified attributions are robust, interpretable, and faithful, enabling reliable use in downstream tasks. Our code is at https://github.com/AlaaAnani/certified-attributions.