Enregistré dans:
Détails bibliographiques
Auteurs principaux: Chen, Tiejin, Huang, Wenwang, Pang, Linsey, Luo, Dongsheng, Wei, Hua
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:https://arxiv.org/abs/2403.06013
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866913260574343168
author Chen, Tiejin
Huang, Wenwang
Pang, Linsey
Luo, Dongsheng
Wei, Hua
author_facet Chen, Tiejin
Huang, Wenwang
Pang, Linsey
Luo, Dongsheng
Wei, Hua
contents This paper delves into the critical area of deep learning robustness, challenging the conventional belief that classification robustness and explanation robustness in image classification systems are inherently correlated. Through a novel evaluation approach leveraging clustering for efficient assessment of explanation robustness, we demonstrate that enhancing explanation robustness does not necessarily flatten the input loss landscape with respect to explanation loss - contrary to flattened loss landscapes indicating better classification robustness. To deeply investigate this contradiction, a groundbreaking training method designed to adjust the loss landscape with respect to explanation loss is proposed. Through the new training method, we uncover that although such adjustments can impact the robustness of explanations, they do not have an influence on the robustness of classification. These findings not only challenge the prevailing assumption of a strong correlation between the two forms of robustness but also pave new pathways for understanding relationship between loss landscape and explanation loss.
format Preprint
id arxiv_https___arxiv_org_abs_2403_06013
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Are Classification Robustness and Explanation Robustness Really Strongly Correlated? An Analysis Through Input Loss Landscape
Chen, Tiejin
Huang, Wenwang
Pang, Linsey
Luo, Dongsheng
Wei, Hua
Machine Learning
Computer Vision and Pattern Recognition
This paper delves into the critical area of deep learning robustness, challenging the conventional belief that classification robustness and explanation robustness in image classification systems are inherently correlated. Through a novel evaluation approach leveraging clustering for efficient assessment of explanation robustness, we demonstrate that enhancing explanation robustness does not necessarily flatten the input loss landscape with respect to explanation loss - contrary to flattened loss landscapes indicating better classification robustness. To deeply investigate this contradiction, a groundbreaking training method designed to adjust the loss landscape with respect to explanation loss is proposed. Through the new training method, we uncover that although such adjustments can impact the robustness of explanations, they do not have an influence on the robustness of classification. These findings not only challenge the prevailing assumption of a strong correlation between the two forms of robustness but also pave new pathways for understanding relationship between loss landscape and explanation loss.
title Are Classification Robustness and Explanation Robustness Really Strongly Correlated? An Analysis Through Input Loss Landscape
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2403.06013