Image Classification on CIFAR-100 (test) (Natural and Robust Accuracy)
64.64Natural AccuracyPGD-2
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PGD-2Backbone=WideResNet-34, Threat Model=L∞, Noise Magnitude=8/2552024.04 | 64.64 | 26.47 | |
| PGD-10Backbone=WideResNet-34, Threat Model=L∞, Noise Magnitude=8/2552024.04 | 60.34 | 30.02 | |
| RS-AAERBackbone=WideResNet-34, Threat Model=L∞, Noise Magnitude=8/2552024.04 | 59.81 | 26.88 | |
| N-FGSMBackbone=WideResNet-34, Threat Model=L∞, Noise Magnitude=8/2552024.04 | 59.02 | 28.49 | |
| N-AAERBackbone=WideResNet-34, Threat Model=L∞, Noise Magnitude=8/2552024.04 | 57.76 | 29.09 | |
| RS-FGSMBackbone=WideResNet-34, Threat Model=L∞, Noise Magnitude=8/2552024.04 | 55.04 | 0 |