Machine Unlearning on ImageNet 1-class unlearning 1K
39.18PGHIU
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| IUBackbone=ResNet182026.03 | 39.18 | 1.1 | 4.15 | 13.71 | 10.85 | 5,684 | |
| BSBackbone=ResNet182026.03 | 35.45 | 0.99 | 1.99 | 8.84 | 7.18 | 44.5 | |
| MUNBaBackbone=ResNet182026.03 | 34.05 | 0.65 | 0 | 11.64 | 8.45 | 16,685.9 | |
| BEBackbone=ResNet182026.03 | 33.94 | 0.92 | 3.96 | 7.17 | 5.79 | 47 | |
| ℓ1-sparseBackbone=ResNet182026.03 | 31.29 | 0.57 | 0 | 7.99 | 5.53 | 11,236.7 | |
| GABackbone=ResNet182026.03 | 30.47 | 0.78 | 7.31 | 3.41 | 2.72 | 24.8 | |
| FTBackbone=ResNet182026.03 | 29.68 | 0.53 | 0.01 | 5.68 | 3.84 | 11,361.4 | |
| SalUnBackbone=ResNet182026.03 | 25.72 | 0.42 | 22.43 | 0.19 | 0.39 | 225,519.2 | |
| DeletionBackbone=ResNet18-KNN2026.03 | 18.99 | 2.54 | 0 | 2.72 | 1.71 | 1 | |
| DeletionBackbone=ResNet18-SPM2026.03 | 14.6 | 0.12 | 0 | 0.19 | 0 | 1 | |
| RetrainBackbone=ResNet182026.03 | 0 | 0 | 0 | 0 | 0 | 226,025.4 | |
| RetrainBackbone=ResNet18-KNN2026.03 | 0 | 0 | 0 | 0 | 0 | 226,025.4 | |
| RetrainBackbone=ResNet18-SPM2026.03 | 0 | 0 | 0 | 0 | 0 | 268,803.5 |