Image Unlearning on Pins Face Recognition Random scenario
8.33Delta Accuracy D'fFT
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| FTBackbone=ViT2026.05 | 8.33 | 1.42 | 60.4 | — | |
| FTBackbone=ResNet182026.05 | 2.78 | -2.6 | 37.7 | — | |
| RLModel=ViT2026.05 | 1.02 | 0.23 | — | 0.01 | |
| SSDModel=ResNet182026.05 | 0.9 | 0.67 | — | -0.06 | |
| LFSSDModel=ResNet182026.05 | 0.87 | 0.72 | — | -0.06 | |
| LFSSDModel=ViT2026.05 | 0.84 | 0.87 | — | 0 | |
| RLModel=ResNet182026.05 | 0.78 | 0.96 | — | 0.06 | |
| BadTModel=ResNet182026.05 | 0.64 | 0.83 | — | -0.01 | |
| SSDModel=ViT2026.05 | 0.63 | 0.61 | — | 0 | |
| BadTModel=ViT2026.05 | 0.36 | 0.16 | — | 0.19 | |
| FTModel=ResNet182026.05 | 0.33 | 0.98 | — | -0.06 | |
| FTModel=ViT2026.05 | 0.32 | 0.23 | — | 0 | |
| BadTBackbone=ViT2026.05 | -8.33 | 0.21 | 32.76 | — | |
| BadTBackbone=ResNet182026.05 | -35 | -32.47 | 41.99 | — | |
| RLBackbone=ResNet182026.05 | -48.89 | -4.59 | 42.11 | — | |
| SSDBackbone=ViT2026.05 | -55 | -54.09 | 171.8 | — | |
| LFSSDBackbone=ResNet182026.05 | -68.33 | -58.61 | 40.82 | — | |
| SSDBackbone=ResNet182026.05 | -75 | -79.97 | 40.25 | — | |
| RLBackbone=ViT2026.05 | -76.67 | 1.33 | 35.71 | — | |
| LFSSDBackbone=ViT2026.05 | -89.44 | -90.7 | 231.91 | — |