Cluster-wise Machine Unlearning on CIFAR100 cluster-wise (Out)
100Cc (Data Retained)∇τ
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| ∇τBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 100 | 24 | 100 | 3 | 95 | 8 | 0.11 | |
| SCRUBBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 100 | 12 | 96 | 14 | 94 | 14 | 0.23 | |
| AMNBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 100 | 92 | 100 | 0 | 95 | 4 | 0.02 | |
| ∇τBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 100 | 24 | 100 | 3 | 95 | 8 | 0.11 | |
| SCRUBBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 100 | 12 | 96 | 14 | 94 | 14 | 0.23 | |
| AMNBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 100 | 92 | 100 | 0 | 95 | 4 | 0.02 | |
| EFFACEBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 99 | 100 | 100 | 82 | 97 | 78 | 0.92 | |
| EFFACEBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 99 | 100 | 100 | 82 | 97 | 78 | 0.92 | |
| SSDBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 96 | 45 | 100 | 36 | 96 | 43 | 0.58 | |
| SSDBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 96 | 45 | 100 | 36 | 96 | 43 | 0.58 | |
| BADTBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 94 | 0 | 100 | 0 | 95 | 1 | 0 | |
| BADTBackbone=RepVGG-A2, c=50, d=50, alpha=0.05, Number of forgotten clusters=52025.08 | 94 | 0 | 100 | 0 | 95 | 1 | 0 |