Machine Unlearning on ImageNet 100 (10% random data forgetting)
27.28UASFRon
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| SFRonBackbone=ViT2025.05 | 27.28 | 78.56 | 77.86 | 61.29 | 21.63 | 88.37 | |
| NegGrad+Backbone=ViT2025.05 | 13.15 | 91.71 | 87.37 | 16.21 | 1.16 | 63.93 | |
| RTBackbone=ViT2025.05 | 11.63 | 91.93 | 87.83 | 13.77 | 0 | 525.72 | |
| MCUβBackbone=ViT2025.05 | 11.63 | 91.92 | 87.7 | 16.21 | 0.65 | 103.52 | |
| MCUBackbone=ViT2025.05 | 11.44 | 92.02 | 87.62 | 16.33 | 0.76 | 103.47 | |
| NegTVBackbone=ViT2025.05 | 10.17 | 91.33 | 87.24 | 12.25 | 1.04 | 11.02 | |
| RLBackbone=ViT2025.05 | 9.53 | 92.06 | 87.82 | 24.32 | 3.2 | 205.73 | |
| SalUnBackbone=ViT2025.05 | 9.38 | 91.94 | 87.73 | 24.29 | 3.22 | 170.34 | |
| GABackbone=ViT2025.05 | 8.96 | 91.15 | 87.53 | 10.5 | 1.76 | 6.71 | |
| FTBackbone=ViT2025.05 | 8.62 | 92.21 | 87.74 | 10.88 | 1.57 | 82.23 |