Scale Matters: Attribution Meets the Wavelet Domain to Explain Model Sensitivity to Image Corruptions - Observation, Impacts, Energie
Pré-Publication, Document De Travail Année : 2023

Scale Matters: Attribution Meets the Wavelet Domain to Explain Model Sensitivity to Image Corruptions

Résumé

Neural networks have shown remarkable performance in computer vision applications, but their deployment in real-world scenarios is challenging due to their sensitivity to image corruptions. Existing attribution methods are uninformative for explaining the sensitivity to image corruptions, while the literature on robustness only provides model-based explanations. However, the ability to scrutinize models' behavior under image corruptions is crucial to increase the user's trust. Towards this end, we introduce the Wavelet sCale Attribution Method (WCAM), a generalization of attribution from the pixel domain to the space-scale domain. Attribution in the space-scale domain reveals where and on what scales the model focuses. We show that the WCAM explains models' failures under image corruptions, identifies sufficient information for prediction, and explain how zoom-in increases accuracy.
Fichier principal
Vignette du fichier
2305.14979.pdf (7.24 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04188020 , version 1 (25-08-2023)

Identifiants

Citer

Gabriel Kasmi, Laurent Dubus, Yves-Marie Saint Drenan, Philippe Blanc. Scale Matters: Attribution Meets the Wavelet Domain to Explain Model Sensitivity to Image Corruptions. 2023. ⟨hal-04188020⟩
120 Consultations
28 Téléchargements

Altmetric

Partager

More