Backdoor Detection on SVHN
100TPRNC
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| NCAttack Type=CB2023.10 | 100 | — | 40 | — | — | — | — | |
| K-ArmAttack Type=BadNet2023.10 | 100 | — | 5 | — | — | — | — | |
| CBDsupAttack Type=BadNet2023.10 | 100 | — | 5 | — | — | — | — | |
| CBDsupAttack Type=CB2023.10 | 100 | — | 5 | — | — | — | — | |
| CBDbeta=0, Attack Type=CB2023.10 | 100 | 100 | 0 | — | — | — | — | |
| CBDbeta=0.1, Attack Type=CB2023.10 | 100 | 100 | 0 | — | — | — | — | |
| CBDbeta=0.2, Attack Type=CB2023.10 | 100 | 100 | 0 | — | — | — | — | |
| NCAttack Type=Blend2023.10 | 95 | — | 40 | — | — | — | — | |
| CBDbeta=0.2, Attack Type=BadNet2023.10 | 95 | 65 | 0 | — | — | — | — | |
| CBDsupAttack Type=Blend2023.10 | 90 | — | 5 | — | — | — | — | |
| CBDbeta=0.1, Attack Type=BadNet2023.10 | 90 | 55 | 0 | — | — | — | — | |
| CBDbeta=0.2, Attack Type=Blend2023.10 | 90 | 80 | 0 | — | — | — | — | |
| FeatureREModel=ConvNet2026.05 | 90 | — | 80 | — | — | — | 10 | |
| UNICORNModel=ResNet-182026.05 | 90 | — | 40 | — | — | — | 50 | |
| NCAttack Type=BadNet2023.10 | 80 | — | 40 | — | — | — | — | |
| CBDbeta=0, Attack Type=Blend2023.10 | 80 | 75 | 0 | — | — | — | — | |
| CBDbeta=0.1, Attack Type=Blend2023.10 | 80 | 80 | 0 | — | — | — | — | |
| CBDbeta=0, Attack Type=BadNet2023.10 | 75 | 45 | 0 | — | — | — | — | |
| K-ArmAttack Type=CB2023.10 | 70 | — | 5 | — | — | — | — | |
| Neural CleanseModel=ViT2026.05 | 60 | — | 40 | — | — | — | 20 | |
| UNICORNModel=ViT2026.05 | 50 | — | 70 | — | — | — | 20 | |
| K-ArmAttack Type=Blend2023.10 | 40 | — | 5 | — | — | — | — | |
| Neural CleanseModel=ResNet-182026.05 | 20 | — | 30 | — | — | — | 10 | |
| MNTDAttack Type=Blend2023.10 | 15 | — | 5 | — | — | — | — | |
| MNTDAttack Type=BadNet2023.10 | 10 | — | 5 | — | — | — | — | |
| MNTDAttack Type=CB2023.10 | 10 | — | 5 | — | — | — | — | |
| Neural CleanseModel=ConvNet2026.05 | 10 | — | 30 | — | — | — | 20 | |
| FeatureREModel=ResNet-182026.05 | 0 | — | 0 | — | — | — | 0 | |
| FeatureREModel=ViT2026.05 | 0 | — | 0 | — | — | — | 0 | |
| UNICORNModel=ConvNet2026.05 | 0 | — | 0 | — | — | — | 0 | |
| Adv-FArchitecture=ResNet2024.08 | — | — | — | — | — | 2.08 | — | |
| Adv-RArchitecture=ResNet2024.08 | — | — | — | — | — | 0.999 | — | |
| DSBAPre-training Dataset=STL-10, Downstream Dataset=SVHN2026.03 | — | — | — | 1.35 | 0.26 | — | — | |
| OriginalArchitecture=ResNet2024.08 | — | — | — | — | — | 1.11 | — | |
| RetrainArchitecture=ResNet2024.08 | — | — | — | — | — | 0.999 | — |