Image Classification on CIFAR-100 (Adversarial Robustness Evaluation)
26.36FGSM AccuracyPoincare ResNet-32
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Poincare ResNet-32Model=PRN 32, ε=2.4/255 (ε1)2025.11 | 26.36 | 21.05 | 12.71 | 10.44 | — | |
| Poincare ResNet-20Model=PRN 20, ε=2.4/255 (ε1)2025.11 | 24.67 | 20.02 | 13.29 | 11.05 | — | |
| Poincare ResNet-32Model=PRN 32, ε=3.2/255 (ε2)2025.11 | 24.61 | 18.74 | 11.18 | 9.19 | 53.44 | |
| Poincare ResNet-20Model=PRN 20, ε=3.2/255 (ε2)2025.11 | 22.62 | 17.66 | 11.68 | 9.28 | 49.63 | |
| Poincare ResNet-32Model=PRN 32, ε=8.0/255 (ε3)2025.11 | 19.67 | 13.93 | 9.24 | 7.86 | — | |
| Poincare ResNet-20Model=PRN 20, ε=8.0/255 (ε3)2025.11 | 17.78 | 12.19 | 9.43 | 9.43 | — |