Adversarial Robustness Evaluation on CIFAR-10 (test)
93.3Accuracy (Natural)NT
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| NTBackbone=WRN-32-102023.03 | 93.3 | 0.01 | 0.02 | 0.05 | 0 | 0 | 0 | 0 | 0.87 | 0.28 | 0 | |
| GeneralistBackbone=WRN-32-102023.03 | 91.03 | 56.88 | 56.92 | 58.87 | 57.23 | 53.94 | 55.8 | 53 | 53.65 | 63.1 | 52.91 | |
| FATBackbone=WRN-32-102023.03 | 89.65 | 48.74 | 48.69 | 48.24 | 52.11 | 48.5 | 48.81 | 46.7 | 46.17 | 51.51 | 44.73 | |
| ATBackbone=WRN-32-10, beta=1/22023.03 | 89.27 | 48.95 | 48.86 | 51.35 | 49.56 | 45.98 | 47.66 | 44.89 | 46.42 | 56.83 | 44.81 | |
| IATBackbone=WRN-32-102023.03 | 87.93 | 50.55 | 50.72 | 52.37 | 48.71 | 47.71 | 46.55 | 43.84 | 45.78 | 56.52 | 43.8 | |
| ATBackbone=WRN-32-10, beta=12023.03 | 87.32 | 49.01 | 48.83 | 48.25 | 52.8 | 48.83 | 49 | 46.34 | 48.17 | 54.26 | 46.11 | |
| RSTBackbone=WRN-32-102023.03 | 87.27 | 46.55 | 46.76 | 47.02 | 45.99 | 45.73 | 46.58 | 45.78 | 43.18 | 52.44 | 41.52 | |
| TRADESBackbone=WRN-32-10, lambda=12023.03 | 87.2 | 51.33 | 51.65 | 52.47 | 53.19 | 51.6 | 51.88 | 49.97 | 50.01 | 54.83 | 49.81 | |
| TRADESBackbone=WRN-32-10, lambda=62023.03 | 85.11 | 54.58 | 54.82 | 55.67 | 54.91 | 54.89 | 55.5 | 52.71 | 52.61 | 57.62 | 52.19 |