Image Classification on Fashion-MNIST Sample Unlearning (Dr)
100AccuracyOriginal
Evaluation results
| Method | Links | |
|---|---|---|
| OriginalBackbone=ResNet182024.11 | 100 | |
| RetrainBackbone=ResNet182024.11 | 100 | |
| NegGradBackbone=ResNet182024.11 | 100 | |
| Residual Feature Alignment UnlearningBackbone=ResNet182024.11 | 100 | |
| FinetuneBackbone=ResNet182024.11 | 99.99 | |
| BadTBackbone=ResNet182024.11 | 99.99 | |
| SCRUBBackbone=ResNet182024.11 | 99.99 | |
| Residual Feature Alignment UnlearningBackbone=ViT2024.11 | 99.95 | |
| RetrainBackbone=ViT2024.11 | 99.88 | |
| FinetuneBackbone=ViT2024.11 | 99.71 | |
| SCRUBBackbone=ViT2024.11 | 99.69 | |
| OriginalBackbone=ViT2024.11 | 99.25 | |
| BadTBackbone=ViT2024.11 | 99.13 | |
| NegGradBackbone=ViT2024.11 | 99.01 |