Machine Unlearning on CIFAR-100 10% unlearning (train)
0Average GapGold Std
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Gold StdBackbone=Vision Transformer (ViT)2025.03 | 0 | 0 | 112.25 | |
| Gold StdBackbone=ResNet18 (RN18)2025.03 | 0 | 0 | 3.39 | |
| LoTUSBackbone=Vision Transformer (ViT)2025.03 | 0.0125 | 0.04 | 7.02 | |
| RndLblBackbone=Vision Transformer (ViT)2025.03 | 0.0175 | 0.06 | 12.79 | |
| SSDBackbone=Vision Transformer (ViT)2025.03 | 0.0175 | 0.04 | 13.67 | |
| SCRUBBackbone=Vision Transformer (ViT)2025.03 | 0.02 | 0.04 | 16.74 | |
| SalUnBackbone=Vision Transformer (ViT)2025.03 | 0.02 | 0.06 | 38.15 | |
| UNSIRBackbone=Vision Transformer (ViT)2025.03 | 0.025 | 0.08 | 10.69 | |
| FinetuningBackbone=Vision Transformer (ViT)2025.03 | 0.0275 | 0.07 | 11.35 | |
| NegGrad+Backbone=Vision Transformer (ViT)2025.03 | 0.0325 | 0.13 | 12.63 | |
| BadTBackbone=Vision Transformer (ViT)2025.03 | 0.0375 | 0.17 | 9.18 | |
| LoTUSBackbone=ResNet18 (RN18)2025.03 | 0.12 | 1.67 | 0.3 | |
| BadTBackbone=ResNet18 (RN18)2025.03 | 0.13 | 4.3 | 0.34 | |
| SCRUBBackbone=ResNet18 (RN18)2025.03 | 0.165 | 3.04 | 0.54 | |
| UNSIRBackbone=ResNet18 (RN18)2025.03 | 0.2125 | 6.5 | 1.55 | |
| NegGrad+Backbone=ResNet18 (RN18)2025.03 | 0.3575 | 6.87 | 0.49 | |
| RndLblBackbone=ResNet18 (RN18)2025.03 | 0.36 | 5.84 | 0.57 | |
| SalUnBackbone=ResNet18 (RN18)2025.03 | 0.3625 | 0.17 | 0.3 | |
| SSDBackbone=ResNet18 (RN18)2025.03 | 0.3675 | 3.05 | 0.45 | |
| FinetuningBackbone=ResNet18 (RN18)2025.03 | 0.4025 | 6.88 | 0.43 |