On Spectral Properties of Gradient-based Explanation Methods

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mehrpanah, Amir, Englesson, Erik, Azizpour, Hossein
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909736905998336
author Mehrpanah, Amir
Englesson, Erik
Azizpour, Hossein
author_facet Mehrpanah, Amir
Englesson, Erik
Azizpour, Hossein
contents Understanding the behavior of deep networks is crucial to increase our confidence in their results. Despite an extensive body of work for explaining their predictions, researchers have faced reliability issues, which can be attributed to insufficient formalism. In our research, we adopt novel probabilistic and spectral perspectives to formally analyze explanation methods. Our study reveals a pervasive spectral bias stemming from the use of gradient, and sheds light on some common design choices that have been discovered experimentally, in particular, the use of squared gradient and input perturbation. We further characterize how the choice of perturbation hyperparameters in explanation methods, such as SmoothGrad, can lead to inconsistent explanations and introduce two remedies based on our proposed formalism: (i) a mechanism to determine a standard perturbation scale, and (ii) an aggregation method which we call SpectralLens. Finally, we substantiate our theoretical results through quantitative evaluations.
format Preprint
id arxiv_https___arxiv_org_abs_2508_10595
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle On Spectral Properties of Gradient-based Explanation Methods
Mehrpanah, Amir
Englesson, Erik
Azizpour, Hossein
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Understanding the behavior of deep networks is crucial to increase our confidence in their results. Despite an extensive body of work for explaining their predictions, researchers have faced reliability issues, which can be attributed to insufficient formalism. In our research, we adopt novel probabilistic and spectral perspectives to formally analyze explanation methods. Our study reveals a pervasive spectral bias stemming from the use of gradient, and sheds light on some common design choices that have been discovered experimentally, in particular, the use of squared gradient and input perturbation. We further characterize how the choice of perturbation hyperparameters in explanation methods, such as SmoothGrad, can lead to inconsistent explanations and introduce two remedies based on our proposed formalism: (i) a mechanism to determine a standard perturbation scale, and (ii) an aggregation method which we call SpectralLens. Finally, we substantiate our theoretical results through quantitative evaluations.
title On Spectral Properties of Gradient-based Explanation Methods
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.10595