Machine Unlearning on CIFAR-100 Random Forget 50%
3.02MIAIU
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| IUBackbone=ResNet-18, Gamma=1/42026.01 | 3.02 | 96.98 | 96.88 | 70.26 | 29.74 | — | — | — | |
| BEBackbone=ResNet-18, Gamma=1/42026.01 | 3.07 | 96.9 | 96.93 | 70.11 | 29.65 | — | — | — | |
| BSBackbone=ResNet-18, Gamma=1/42026.01 | 3.08 | 96.92 | 96.88 | 70.3 | 29.71 | — | — | — | |
| GABackbone=ResNet-18, Gamma=1/42026.01 | 3.44 | 96.56 | 96.86 | 69.1 | 29.24 | — | — | — | |
| FTBackbone=ResNet-18, Gamma=1/42026.01 | 5.78 | 95.79 | 99.96 | 69.15 | 27.71 | — | — | — | |
| l1-sparseBackbone=ResNet-18, Gamma=1/42026.01 | 24.68 | 75.32 | 88.76 | 61.54 | 18.76 | — | — | — | |
| RLBackbone=ResNet-18, Gamma=1/42026.01 | 28.41 | 67.78 | 96.3 | 54.35 | 12.26 | — | — | — | |
| SalUnBackbone=ResNet-18, Gamma=1/42026.01 | 48.15 | 52.59 | 96.12 | 54.71 | 3.66 | — | — | — | |
| RetrainBackbone=ResNet-18, Gamma=1/42026.01 | 51.56 | 48.44 | 99.94 | 51.46 | 0 | — | — | — | |
| FalWBackbone=ResNet-18, Gamma=1/42026.01 | 52.94 | 48.06 | 99.98 | 52.05 | 0.6 | — | — | — | |
| SFRonBackbone=ResNet-18, Gamma=1/42026.01 | 53.7 | 46.3 | 99.98 | 54.78 | 1.91 | — | — | — | |
| FTBackbone=ViT2026.04 | 72 | — | — | — | 0.129 | 99.9 | 91.8 | 58 | |
| IEU w/GABackbone=ViT2026.04 | 72.3 | — | — | — | 0.098 | 99.8 | 81.7 | 55.9 | |
| SCRUBBackbone=ViT2026.04 | 72.6 | — | — | — | 0.117 | 99.8 | 86 | 59.6 | |
| RetrainingBackbone=ViT2026.04 | 74.1 | — | — | — | 0 | 100 | 49.7 | 50.6 | |
| IEU w/GA+NoisyBackbone=ViT2026.04 | 74.2 | — | — | — | 0.013 | 97.3 | 52.1 | 50.7 | |
| IEU w/NoisyBackbone=ViT2026.04 | 75.1 | — | — | — | 0.016 | 97.2 | 52.3 | 50.5 | |
| SALUNBackbone=ViT2026.04 | 76.1 | — | — | — | 0.145 | 54.8 | 45.1 | 44.4 | |
| RLBackbone=ViT2026.04 | 80.3 | — | — | — | 0.037 | 99.8 | 55.5 | 47.9 |