Machine Unlearning on CIFAR-100 subclass-wise
100Accuracy (Deleted Data)Original
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| OriginalDr=✓, Df=✓2025.08 | 100 | 100 | 77.8 | 0 | — | |
| RetrainDr=✓, Df=✗2025.08 | 0 | 99.33 | 76.49 | 1 | 0 | |
| IU*Dr=✓, Df=✓, Run time (s)=18±02025.08 | 0 | 60.58 | 54.5 | 1 | 5.8 | |
| GADr=✗, Df=✓, Run time (s)=9±02025.08 | 0 | 75.56 | 49.5 | 1 | 7.54 | |
| RLDr=✗, Df=✓, Run time (s)=19±22025.08 | 0 | 76.49 | 47.69 | 1 | 7.57 | |
| SSDDr=✓, Df=✓, Run time (s)=10±02025.08 | 0 | 83.07 | 52.2 | 1 | 4.81 | |
| SCAR*Dr=✗, Df=✓, Run time (s)=67±02025.08 | 0 | 78.72 | 56.55 | 1 | 5.88 | |
| NPODr=✗, Df=✓, Run time (s)=26±22025.08 | 0 | 90.22 | 58.9 | 1 | 5.61 | |
| IMUDr=✗, Df=✓, Run time (s)=14±02025.08 | 0 | 98.61 | 67.6 | 1 | 3.69 |