Is RobustBench/AutoAttack a suitable Benchmark for Adversarial Robustness?

arXiv:2112.01601v4 Announce Type: replace-cross
Abstract: 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 …

adversarial arxiv benchmark classification cs.cr cs.cv image networks neural networks robustness suitable task under