Machine Unlearning on CIFAR-100 Forget 50%
75.69Test AccuracyBASE
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| BASEBackbone=ResNet-182026.03 | 75.69 | 67.83 | 15.38 | |
| SSDBackbone=ResNet-182026.03 | 75.68 | 67.83 | 15.37 | |
| NEGGRADBackbone=ResNet-182026.03 | 75.33 | 67.94 | 15.31 | |
| RETRAINBackbone=ResNet-182026.03 | 65.77 | 50.43 | 0 | |
| ℓ1-SPARSEBackbone=ResNet-182026.03 | 62.6 | 58.05 | 5.58 | |
| AMUNBackbone=ResNet-182026.03 | 62.45 | 55.05 | 4.04 | |
| FINETUNEBackbone=ResNet-182026.03 | 61.64 | 59.43 | 6.43 | |
| REGUNBackbone=ResNet-182026.03 | 61.55 | 55.16 | 3.51 | |
| NEGGRAD+Backbone=ResNet-182026.03 | 60.85 | 56.68 | 5.21 | |
| SALUNBackbone=ResNet-182026.03 | 59.02 | 56.53 | 5.19 |