Image Classification on CIFAR-10 (test) (BA and ASR Metrics)
86.7Base Accuracy (BA)No defense
Evaluation results
| Method | Links | ||
|---|---|---|---|
| No defenseAttack=BadNets, Architecture=12-layer Swin transformer2024.05 | 86.7 | 97.43 | |
| No defenseAttack=Blend, Architecture=12-layer Swin transformer2024.05 | 85.46 | 100 | |
| BTI-DBF(U)Attack=BadNets, Architecture=12-layer Swin transformer2024.05 | 85.21 | 99.99 | |
| BAN Fine-tuneAttack=BadNets, Architecture=12-layer Swin transformer2024.05 | 83.92 | 2.91 | |
| BTI-DBF(U)Attack=Blend, Architecture=12-layer Swin transformer2024.05 | 83.9 | 100 | |
| FTAttack=BadNets, Architecture=12-layer Swin transformer2024.05 | 82.23 | 44.07 | |
| FTAttack=Blend, Architecture=12-layer Swin transformer2024.05 | 80.13 | 100 | |
| No defenseAttack=Bpp, Architecture=12-layer Swin transformer2024.05 | 79.19 | 63.74 | |
| BAN Fine-tuneAttack=Blend, Architecture=12-layer Swin transformer2024.05 | 79.18 | 26.2 | |
| No defenseAttack=WaNet, Architecture=12-layer Swin transformer2024.05 | 78.25 | 31.03 | |
| No defenseAttack=IAD, Architecture=12-layer Swin transformer2024.05 | 76.35 | 88.71 | |
| BAN Fine-tuneAttack=IAD, Architecture=12-layer Swin transformer2024.05 | 76.26 | 9.8 | |
| FTAttack=WaNet, Architecture=12-layer Swin transformer2024.05 | 75.63 | 2.3 | |
| FTAttack=Bpp, Architecture=12-layer Swin transformer2024.05 | 75.35 | 12.99 | |
| BAN Fine-tuneAttack=WaNet, Architecture=12-layer Swin transformer2024.05 | 75.28 | 3.1 | |
| BTI-DBF(U)Attack=Bpp, Architecture=12-layer Swin transformer2024.05 | 74.66 | 11.04 | |
| FTAttack=IAD, Architecture=12-layer Swin transformer2024.05 | 74.58 | 54.63 | |
| BTI-DBF(U)Attack=IAD, Architecture=12-layer Swin transformer2024.05 | 74.29 | 61.16 | |
| BAN Fine-tuneAttack=Bpp, Architecture=12-layer Swin transformer2024.05 | 74.25 | 6.91 | |
| BTI-DBF(U)Attack=WaNet, Architecture=12-layer Swin transformer2024.05 | 73.83 | 2.61 |