Image Classification on CIFAR-10 (test) (Adversarial Robustness & Speed)
87.05Accuracy (Clean)AT
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ATBackbone=ResNet-18, Sparsity rate=0%, Params=11.2M2025.04 | 87.05 | 56.14 | 48.02 | 57.6 | 53.1 | 1 | |
| EEDBackbone=ResNet-18, Sparsity rate=80%, Params=2.2M2025.04 | 86.13 | 55.71 | 48.13 | 57.03 | 51.97 | 1.86 | |
| HARPBackbone=ResNet-18, Sparsity rate=80%, Params=2.2M2025.04 | 83.84 | 52.56 | 45.36 | 56.57 | 47.04 | 1.81 | |
| ATBackbone=VGG-16, Sparsity rate=0%, Params=14.7M2025.04 | 82.7 | 54.49 | 48.52 | 54.91 | 56.91 | 1 | |
| EEDBackbone=VGG-16, Sparsity rate=80%, Params=2.9M2025.04 | 81.39 | 54.26 | 47.49 | 53.27 | 53.94 | 1.79 | |
| R-ADMMBackbone=ResNet-18, Sparsity rate=80%, Params=2.2M2025.04 | 81.25 | 48 | 43.92 | 49.17 | 39.11 | 1.68 | |
| Flying BirdBackbone=ResNet-18, Sparsity rate=80%, Params=2.2M2025.04 | 81.07 | 51.62 | 44.41 | 56.08 | 45.29 | 1.79 | |
| HARPBackbone=VGG-16, Sparsity rate=80%, Params=2.9M2025.04 | 80.92 | 51.77 | 44.02 | 52.79 | 52.17 | 1.73 | |
| HYDRABackbone=VGG-16, Sparsity rate=80%, Params=2.9M2025.04 | 78.83 | 47.85 | 44.85 | 45.73 | 49.41 | 1.72 | |
| HYDRABackbone=ResNet-18, Sparsity rate=80%, Params=2.2M2025.04 | 77.36 | 52.92 | 43.74 | 49.64 | 45.91 | 1.77 | |
| Flying BirdBackbone=VGG-16, Sparsity rate=80%, Params=2.9M2025.04 | 77.36 | 49.43 | 45.73 | 50.38 | 48.34 | 1.68 | |
| RSTBackbone=VGG-16, Sparsity rate=80%, Params=2.9M2025.04 | 77.29 | 39.9 | 30.15 | 32.02 | 42.03 | 1.83 | |
| TwinRepBackbone=ResNet-18, Sparsity rate=80%, Params=2.2M2025.04 | 77.26 | 52.04 | 43.52 | 56.6 | 46.18 | 1.82 | |
| R-ADMMBackbone=VGG-16, Sparsity rate=80%, Params=2.9M2025.04 | 76.51 | 47.19 | 40.41 | 44.8 | 48.29 | 1.65 | |
| TwinRepBackbone=VGG-16, Sparsity rate=80%, Params=2.9M2025.04 | 76.04 | 50.12 | 43.26 | 52.42 | 51.59 | 1.79 | |
| MADBackbone=ResNet-18, Sparsity rate=80%, Params=2.2M2025.04 | 74.18 | 50.38 | 41.27 | 54.17 | 38.44 | 1.69 | |
| MADBackbone=VGG-16, Sparsity rate=80%, Params=2.9M2025.04 | 73.65 | 48.81 | 40.88 | 47.16 | 45.5 | 1.8 | |
| RSTBackbone=ResNet-18, Sparsity rate=80%, Params=2.2M2025.04 | 61.02 | 41.01 | 18.38 | 51.02 | 26.82 | 1.86 |