Backdoor Defense on Tiny-ImageNet Blended (test)
73.88Clean AccuracyClean
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CleanMode=O-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 73.88 | 0.1 | |
| AttackMode=O-Backdoor, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 73.68 | 99.99 | |
| CleanMode=P-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 73.48 | 0.6 | |
| SAMMode=P-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 71.3 | 99.41 | |
| EPMode=O-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 71.11 | 0.01 | |
| EPMode=P-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 70.93 | 0.09 | |
| SAMMode=O-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 70.8 | 4.02 | |
| BTIMode=O-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 68.74 | 0.55 | |
| BTIMode=P-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 68.41 | 38.13 | |
| PAMMode=O-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 67.89 | 0.14 | |
| PAMMode=P-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 66.86 | 0.37 | |
| FSTMode=O-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 66.55 | 1.19 | |
| FSTMode=P-Robustness, Backbone=ResNet-18, Poisoning Rate=5%2024.10 | 65.58 | 60.12 |