Machine Unlearning on CIFAR-10 Forget 1%
94.34Test AccuracyRETRAIN
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| RETRAINBackbone=ResNet-182026.03 | 94.34 | 49.98 | 0 | |
| BASEBackbone=ResNet-182026.03 | 94.2 | 60.11 | 3.88 | |
| NEGGRADBackbone=ResNet-182026.03 | 94.17 | 59.8 | 3.82 | |
| SSDBackbone=ResNet-182026.03 | 93.82 | 59.69 | 3.84 | |
| AMUNBackbone=ResNet-182026.03 | 91.84 | 44.17 | 3.94 | |
| NEGGRAD+Backbone=ResNet-182026.03 | 91.8 | 57.95 | 3.77 | |
| SALUNBackbone=ResNet-182026.03 | 91.63 | 50.09 | 1.64 | |
| ℓ1-SPARSEBackbone=ResNet-182026.03 | 90.97 | 53.89 | 2.73 | |
| REGUNBackbone=ResNet-182026.03 | 90.93 | 48.9 | 1.21 | |
| FINETUNEBackbone=ResNet-182026.03 | 90.9 | 54.78 | 2.88 |