Machine Unlearning on PathMNIST External (test)
88.63Residual Accuracy (Acc_r)Retrained Model
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Retrained ModelBackbone=ViT, Data requirement=Retraining (classifier head only)2025.11 | 88.63 | 0 | 1.02 | — | — | |
| POUR-PBackbone=ViT, Data requirement=Forget set only2025.11 | 87.44 | 0 | 1 | — | 84.2 | |
| Original ModelBackbone=ViT, Data requirement=Full Dataset2025.11 | 87.13 | 97.53 | 0.51 | — | 84.2 | |
| FinetuneBackbone=ViT, Data requirement=Requires retain set2025.11 | 86.99 | 0 | 1 | 6 | 86.1 | |
| FCSBackbone=ViT, Data requirement=Requires retain set2025.11 | 83.86 | 0 | 0.97 | 10 | 96.6 | |
| POUR-DBackbone=ViT, Data requirement=Forget set only2025.11 | 80.9 | 7.92 | 0.87 | 61 | 85.2 | |
| Gradient AscentBackbone=ViT, Data requirement=Forget set only2025.11 | 76.72 | 10.61 | 0.81 | 26 | 83.9 | |
| DELETEBackbone=ViT, Data requirement=Forget set only2025.11 | 73.04 | 0 | 0.86 | 42 | 88.4 | |
| Random LabelBackbone=ViT, Data requirement=Forget set only2025.11 | 70.61 | 25.34 | 0.67 | 29 | 83.6 |