Image Classification on CIFAR-10 (test) (Adversarial Robustness: PGD 10/100, C&W, AA)
87.54Accuracy (Clean)RAAT
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| RAATBackbone=WideResNet-28-102026.04 | 87.54 | 58.22 | 57.64 | 54.68 | 51.76 | |
| PGD-ATBackbone=WideResNet-28-102026.04 | 87.49 | 55.81 | 54.37 | 52.03 | 50.96 | |
| Cons-ATBackbone=WideResNet-28-102026.04 | 86.9 | 55.71 | 54.41 | 52.95 | 50.83 | |
| TRADESBackbone=WideResNet-28-102026.04 | 85.35 | 57.15 | 56.17 | 53.61 | 51.88 | |
| RAATBackbone=ResNet-182026.04 | 83.69 | 54.55 | 53.5 | 51.07 | 47.94 | |
| Cons-ATBackbone=ResNet-182026.04 | 83.42 | 53.2 | 51.68 | 49.37 | 47.72 | |
| PGD-ATBackbone=ResNet-182026.04 | 82.92 | 51.81 | 50.34 | 48.51 | 46.74 | |
| RAAT++Backbone=WideResNet-28-102026.04 | 82.8 | 58.76 | 57.88 | 55.42 | 52.1 | |
| MARTBackbone=WideResNet-28-102026.04 | 82.78 | 58.47 | 57.47 | 53.21 | 50.89 | |
| RAAT++Backbone=ResNet-182026.04 | 79.7 | 54.87 | 53.87 | 51.64 | 47.5 | |
| TRADESBackbone=ResNet-182026.04 | 79.67 | 52.14 | 51.88 | 49.33 | 47.62 | |
| MARTBackbone=ResNet-182026.04 | 77.93 | 53.61 | 52.83 | 48.66 | 46.7 |