Machine Unlearning on CIFAR-10 10% random unlearning
0PGHRetrain
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| RetrainBackbone=ResNet182026.03 | 0 | 0 | 0 | 5.35 | 0 | 100 | 0 | 94.25 | |
| RetrainBackbone=ResNet18-KNN2026.03 | 0 | 0 | 0 | 0.02 | 0 | 99.98 | 0 | 94.45 | |
| RetrainBackbone=ResNet18-SPM2026.03 | 0 | 0 | 0 | 0.18 | 0 | 99.43 | 0 | 93.89 | |
| GABackbone=ResNet182026.03 | 5.23 | 0.17 | 4.8 | 0.55 | 4.8 | 99.52 | 0.48 | 94.57 | |
| BEBackbone=ResNet182026.03 | 5.3 | 0.17 | 4.83 | 0.52 | 0.48 | 99.52 | 0.29 | 94.54 | |
| BSBackbone=ResNet182026.03 | 5.31 | 0.17 | 4.82 | 0.53 | 0.49 | 99.51 | 0.28 | 94.53 | |
| MUNBaBackbone=ResNet182026.03 | 5.39 | 0.14 | 4.39 | 0.96 | 0.03 | 99.97 | 0.19 | 94.44 | |
| SalUnBackbone=ResNet182026.03 | 5.4 | 0.15 | 3.89 | 1.46 | 0.14 | 99.86 | 0.08 | 94.17 | |
| SPMBackbone=ResNet18-SPM2026.03 | 5.54 | 0.19 | 0.16 | 0.02 | 0.55 | 99.98 | 0.56 | 94.45 | |
| FTBackbone=ResNet182026.03 | 5.89 | 0.19 | 4.46 | 0.89 | 0.16 | 99.84 | 0.32 | 93.93 | |
| l1-sparseBackbone=ResNet182026.03 | 6.03 | 0.19 | 4.29 | 1.06 | 0.24 | 99.76 | 0.5 | 93.75 | |
| IUBackbone=ResNet182026.03 | 26.6 | 1.69 | 17.15 | 22.5 | 22.32 | 77.68 | 21.09 | 73.16 |