Multi-class Classification on Imagenet Small (test)
8.8Average Prediction Set SizePCS Original
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PCS OriginalBackbone=Res-net 18, Target coverage=90%, Number of bootstraps (B)=1002025.05 | 8.8 | 500 | |
| PCS NoiseBackbone=Res-net 18, Target coverage=90%, Number of bootstraps (B)=10002025.05 | 9.6 | 5 | |
| PCS DropoutBackbone=Res-net 18, Target coverage=90%, Number of bootstraps (B)=10002025.05 | 9.8 | 5 | |
| RAPSBackbone=Res-net 18, Target coverage=90%2025.05 | 11.2 | 3 | |
| TopKBackbone=Res-net 18, Target coverage=90%2025.05 | 13 | 3 | |
| APSBackbone=Res-net 18, Target coverage=90%2025.05 | 14.4 | 3 |