Spectral Integrated Gradients for Coarse-to-Fine Feature Attribution

TL;DR

Spectral Integrated Gradients (SIG) enhances feature attribution by using SVD-based paths for a coarse-to-fine explanation, reducing noise and improving interpretability in image classification.

Contribution

This paper introduces Spectral Integrated Gradients, a novel path construction method based on SVD that improves attribution quality over traditional straight-line paths.

Findings

SIG produces cleaner attribution maps with less noise.

SIG achieves better quantitative performance than existing methods.

Extensive evaluations confirm the effectiveness of SIG across datasets.

Abstract

Integrated Gradients (IG) is a widely adopted feature attribution method that satisfies desirable axiomatic properties. However, the choice of integration path significantly affects the quality of attributions, and the standard straight-line path introduces all input features simultaneously, often accumulating noisy gradients along the way. To address this limitation, we propose Spectral Integrated Gradients, which constructs integration paths based on singular value decomposition (SVD) of the baseline-to-input difference. By progressively activating singular components from largest to smallest, SIG introduces global structure before fine-grained details, naturally following a coarse-to-fine progression. Through extensive evaluation across diverse image classification datasets, we demonstrate that SIG produces cleaner attribution maps with reduced noise and achieves improved quantitative performance compared to existing path-based attribution methods. Our code is available at https://github.com/leekwoon/sig/.