ResearchBenchmarksImage Classification on CIFAR-10 long-tail (IR=20) (test) (Adversarial Robustness)For AI assistantsMarkdownContent indexImage Classification on CIFAR-10 long-tail (IR=20) (test) (Adversarial Robustness)Follow78.49Accuracy (Clean)RoBal64.637268.233671.8375.4264Apr 6, 2021Nov 29, 2021Jul 25, 2022Mar 20, 2023Nov 12, 2023Jul 7, 2024Mar 2, 2025Evaluation resultsMethodMethodLinksAccuracy (Clean)Accuracy (FGSM)Accuracy (PGD-20)Accuracy (CW)Accuracy (AA)Accuracy (MIM)RoBal2021.0478.49—39.1739.3837.5840.44AT2021.0474.09—33.5934.2732.0234.65TAETBackbone=ResNet-18, IR=20Backbone=ResNet-18, IR=202025.0366.9336.8233.9858.4127.84—AT-BSLBackbone=ResNet-18, IR=20Backbone=ResNet-18, IR=202025.0366.428.8222.5615.4321.46—TRADES2021.0465.17—34.6534.0733.0635.29