Is RobustBench/AutoAttack a suitable Benchmark for Adversarial Robustness?

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lorenz, Peter, Strassel, Dominik, Keuper, Margret, Keuper, Janis
Format: Preprint
Published: 2021
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929248847003648
author Lorenz, Peter
Strassel, Dominik
Keuper, Margret
Keuper, Janis
author_facet Lorenz, Peter
Strassel, Dominik
Keuper, Margret
Keuper, Janis
contents Recently, RobustBench (Croce et al. 2020) has become a widely recognized benchmark for the adversarial robustness of image classification networks. In its most commonly reported sub-task, RobustBench evaluates and ranks the adversarial robustness of trained neural networks on CIFAR10 under AutoAttack (Croce and Hein 2020b) with l-inf perturbations limited to eps = 8/255. With leading scores of the currently best performing models of around 60% of the baseline, it is fair to characterize this benchmark to be quite challenging. Despite its general acceptance in recent literature, we aim to foster discussion about the suitability of RobustBench as a key indicator for robustness which could be generalized to practical applications. Our line of argumentation against this is two-fold and supported by excessive experiments presented in this paper: We argue that I) the alternation of data by AutoAttack with l-inf, eps = 8/255 is unrealistically strong, resulting in close to perfect detection rates of adversarial samples even by simple detection algorithms and human observers. We also show that other attack methods are much harder to detect while achieving similar success rates. II) That results on low-resolution data sets like CIFAR10 do not generalize well to higher resolution images as gradient-based attacks appear to become even more detectable with increasing resolutions.
format Preprint
id arxiv_https___arxiv_org_abs_2112_01601
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle Is RobustBench/AutoAttack a suitable Benchmark for Adversarial Robustness?
Lorenz, Peter
Strassel, Dominik
Keuper, Margret
Keuper, Janis
Computer Vision and Pattern Recognition
Cryptography and Security
Recently, RobustBench (Croce et al. 2020) has become a widely recognized benchmark for the adversarial robustness of image classification networks. In its most commonly reported sub-task, RobustBench evaluates and ranks the adversarial robustness of trained neural networks on CIFAR10 under AutoAttack (Croce and Hein 2020b) with l-inf perturbations limited to eps = 8/255. With leading scores of the currently best performing models of around 60% of the baseline, it is fair to characterize this benchmark to be quite challenging. Despite its general acceptance in recent literature, we aim to foster discussion about the suitability of RobustBench as a key indicator for robustness which could be generalized to practical applications. Our line of argumentation against this is two-fold and supported by excessive experiments presented in this paper: We argue that I) the alternation of data by AutoAttack with l-inf, eps = 8/255 is unrealistically strong, resulting in close to perfect detection rates of adversarial samples even by simple detection algorithms and human observers. We also show that other attack methods are much harder to detect while achieving similar success rates. II) That results on low-resolution data sets like CIFAR10 do not generalize well to higher resolution images as gradient-based attacks appear to become even more detectable with increasing resolutions.
title Is RobustBench/AutoAttack a suitable Benchmark for Adversarial Robustness?
topic Computer Vision and Pattern Recognition
Cryptography and Security
url https://arxiv.org/abs/2112.01601