Expected Grad-CAM: Towards gradient faithfulness

TL;DR

Expected Grad-CAM introduces a novel gradient-weighted CAM method that enhances faithfulness and robustness by reshaping gradient computation using Expected Gradients and kernel smoothing, addressing saturation and sensitivity issues.

Contribution

The paper proposes Expected Grad-CAM, a new gradient-weighted CAM approach that improves explanation faithfulness and robustness by integrating well-established gradient smoothing techniques.

Findings

Improves explanation faithfulness and robustness.

Effectively mitigates saturation and sensitivity issues.

Provides both quantitative and qualitative validation.

Abstract

Although input-gradients techniques have evolved to mitigate and tackle the challenges associated with gradients, modern gradient-weighted CAM approaches still rely on vanilla gradients, which are inherently susceptible to the saturation phenomena. Despite recent enhancements have incorporated counterfactual gradient strategies as a mitigating measure, these local explanation techniques still exhibit a lack of sensitivity to their baseline parameter. Our work proposes a gradient-weighted CAM augmentation that tackles both the saturation and sensitivity problem by reshaping the gradient computation, incorporating two well-established and provably approaches: Expected Gradients and kernel smoothing. By revisiting the original formulation as the smoothed expectation of the perturbed integrated gradients, one can concurrently construct more faithful, localized and robust explanations which minimize infidelity. Through fine modulation of the perturbation distribution it is possible to regulate the complexity characteristic of the explanation, selectively discriminating stable features. Our technique, Expected Grad-CAM, differently from recent works, exclusively optimizes the gradient computation, purposefully designed as an enhanced substitute of the foundational Grad-CAM algorithm and any method built therefrom. Quantitative and qualitative evaluations have been conducted to assess the effectiveness of our method.