Image Classification on CIFAR10 (test) (Clean & Certified Accuracy)
93.48Clean AccuracyECViT-B
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ECViT-BDefense Category=Smooth Model, Backbone=ViT-B2022.03 | 93.48 | 82.8 | 76.38 | |
| ViT-BDefense Category=Smooth Model2022.03 | 91.28 | 78.1 | 70.78 | |
| ECViT-SDefense Category=Smooth Model, Backbone=ViT-S2022.03 | 87.56 | 73.82 | 65.1 | |
| BagCertDefense Category=Baseline2022.03 | 86 | 73.33 | 64.9 | |
| ResNext101Defense Category=Smooth Model2022.03 | 85.34 | 69.32 | 60.68 | |
| PGDefense Category=Baseline2022.03 | 84.7 | 69.2 | 57.7 | |
| CBNDefense Category=Baseline2022.03 | 84.2 | 44.2 | 9.3 | |
| DSDefense Category=Baseline2022.03 | 83.9 | 68.9 | 56.2 | |
| ViT-SDefense Category=Smooth Model2022.03 | 80.4 | 61.5 | 51.78 |