Backdoor Defense on CIFAR-10 LC DenseNet-161 (test)
90.17Clean AccuracyClean
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CleanMode=O-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 90.17 | 2.17 | |
| AttackMode=O-Backdoor, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 90.13 | 99.8 | |
| ANPMode=O-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 90.04 | 1.58 | |
| ANPMode=P-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 89.45 | 71.1 | |
| CleanMode=P-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 89.41 | 4.44 | |
| EPMode=P-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 88.94 | 0.95 | |
| BAUMode=P-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 88.84 | 82.47 | |
| EPMode=O-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 88.59 | 0.56 | |
| BAUMode=O-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 88.02 | 2.63 | |
| BTIMode=O-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 87.93 | 2.71 | |
| SAMMode=O-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 87.79 | 2.52 | |
| PAMMode=O-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 87.7 | 1.81 | |
| BTIMode=P-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 87.67 | 100 | |
| FSTMode=P-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 87.47 | 72.61 | |
| FSTMode=O-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 87.4 | 1.57 | |
| SAMMode=P-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 87.3 | 12.23 | |
| PAMMode=P-Robustness, Backbone=DenseNet-161, Poisoning Rate=5%2024.10 | 86.44 | 9.39 |