Random Subset Unlearning on SVHN
100Retention Accuracy (RA)Pretrain
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| PretrainBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 100 | 100 | 96.48 | 1.65 | 83.11 | 16,904 | |
| RetrainBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 100 | 95.12 | 96.4 | 0 | 81.24 | 13,890 | |
| FTBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 100 | 96.38 | 96.87 | 0.54 | 80.72 | 805 | |
| SalUnBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 99.77 | 94.5 | 95.85 | 0.47 | 19.92 | 409 | |
| MunBaBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 99.73 | 96.46 | 96.84 | 0.68 | 66.7 | 583 | |
| NGBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 98.98 | 98.98 | 96.56 | 1.53 | 84.53 | 25 | |
| BTBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 98.73 | 96.71 | 95.01 | 1.42 | 22.32 | 215 | |
| SSDBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 98.67 | 98.63 | 96.59 | 1.52 | 84.23 | 55 | |
| RLBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 98.66 | 98.53 | 96.02 | 1.56 | 73.94 | 25 | |
| SCARBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 96.34 | 96.46 | 92.39 | 2.86 | 55.2 | 158 | |
| JiTBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 96.3 | 96.26 | 95.14 | 1.88 | 62.78 | 333 | |
| DUCKBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 96.16 | 94.86 | 95.53 | 1.51 | 30.46 | 156 | |
| MU-MisBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 95.48 | 95.5 | 94.22 | 2.36 | 26.11 | 116 | |
| SCRUBBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 95.4 | 94.79 | 94.94 | 1.98 | 23.15 | 104 | |
| LoTusBackbone=ResNet-18, Unlearning ratio=10%, Task Type=Random subset unlearning2024.02 | 94.48 | 94.31 | 86.65 | 5.36 | 78.8 | 35 |