Machine Unlearning on ImageNet100 30% influence-style deletion (retain)
71.8Retention AccuracyORTR
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ORTRBackbone=ResNet-50, Deletion Setting=30% influence-style deletion, Seeds=52026.02 | 71.8 | 0.533 | 0 | 0 | |
| SRAGUBackbone=ResNet-50, Deletion Setting=30% influence-style deletion, Seeds=52026.02 | 42.99 | 0.705 | 0.122 | 0.222 | |
| BoundBackbone=ResNet-50, Deletion Setting=30% influence-style deletion, Seeds=52026.02 | 42.86 | 0.835 | 0.438 | 0.557 | |
| SalUnBackbone=ResNet-50, Deletion Setting=30% influence-style deletion, Seeds=52026.02 | 42.2 | 0.812 | 0.412 | 0.518 | |
| AmnBackbone=ResNet-50, Deletion Setting=30% influence-style deletion, Seeds=52026.02 | 42.05 | 0.792 | 0.352 | 0.428 | |
| SISABackbone=ResNet-50, Deletion Setting=30% influence-style deletion, Seeds=52026.02 | 41.9 | 0.758 | 0.368 | 0.452 | |
| AGUBackbone=ResNet-50, Deletion Setting=30% influence-style deletion, Seeds=52026.02 | 40.85 | 0.862 | 0.468 | 0.599 | |
| SCRUBBackbone=ResNet-50, Deletion Setting=30% influence-style deletion, Seeds=52026.02 | 40.45 | 0.778 | 0.385 | 0.478 |