List your company
ResearchFormerly WizWand

Image Classification on ImageNet (Standard and Robust Accuracy)

82.33Standard AccuracyWithout defense
73.17875.55477.9380.306Aug 30, 2024
Updated 8mo ago

Evaluation results

MethodLinks
Without defenseAttack Method=random-targeted PGD-40, Architecture=ResNet-1522024.0882.330.04
Without defenseAttack Method=untargeted PGD-100, Architecture=ResNet-502024.0880.550.01
Without defenseAttack Method=AutoAttack, Architecture=ResNet-502024.0880.550
OSCPDefense=Hybrid, Attack Method=random-targeted PGD-40, Architecture=ResNet-1522024.0879.8178.78
Wang et al.Defense=Diffusion Based Purification, Attack Method=random-targeted PGD-40, Architecture=ResNet-1522024.0878.177.86
Amini et al.Defense=Adversarial Training, Attack Method=AutoAttack, Architecture=ConvNeXt-L2024.0877.9659.64
OSCPDefense=Hybrid, Attack Method=untargeted PGD-100, Architecture=ResNet-502024.0877.6373.89
OSCPDefense=Hybrid, Attack Method=AutoAttack, Architecture=ResNet-502024.0877.6374.19
Singh et al.Defense=Adversarial Training, Attack Method=AutoAttack, Architecture=ConvNeXt-L2024.087757.7
DiffPureDefense=Diffusion Based Purification, Attack Method=AutoAttack, Architecture=ResNet-502024.0875.7773.02
Wang et al.Defense=Diffusion Based Purification, Attack Method=untargeted PGD-100, Architecture=ResNet-502024.0873.5372.97