Image Classification on CIFAR100 8/255 noise (test)
60.29Natural AccuracyRS-FGSM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RS-FGSMBackbone=PreActResNet-182024.04 | 60.29 | 10.58 | |
| PGD-2Backbone=PreActResNet-182024.04 | 60.09 | 24.58 | |
| RS-AAERBackbone=PreActResNet-182024.04 | 57.71 | 25.31 | |
| PGD-10(20)Backbone=PreActResNet-182024.04 | 55.05 | 27.87 | |
| N-FGSMBackbone=PreActResNet-182024.04 | 54.92 | 26.29 | |
| N-AAERBackbone=PreActResNet-182024.04 | 54.47 | 26.81 | |
| Grad AlignBackbone=PreActResNet-182024.04 | 54 | 25.27 | |
| ZeroGradBackbone=PreActResNet-182024.04 | 53.83 | 25.02 | |
| MultiGradBackbone=PreActResNet-182024.04 | 53.11 | 25.68 | |
| FreeATBackbone=PreActResNet-182024.04 | 47.41 | 22.27 |