Misclassification Detection on HAM10000
85.5AUCCUPID Epistemic
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CUPID EpistemicBackbone=ResNet18, Uncertainty Estimation Type=Epistemic2026.03 | 85.5 | 4.7 | 0.907 | |
| Rate-inBackbone=ResNet182026.03 | 84.6 | 4.8 | 0.915 | |
| DECBackbone=ResNet182026.03 | 83.7 | 8.2 | 0.874 | |
| MC DropoutBackbone=ResNet182026.03 | 82.9 | 7.6 | 0.861 | |
| PostNet AleatoricBackbone=ResNet18, Uncertainty Estimation Type=Aleatoric2026.03 | 79.3 | 14.2 | 0.764 | |
| BNNBackbone=ResNet182026.03 | 79.3 | 9.6 | 0.821 | |
| CUPID AleatoricBackbone=ResNet18, Uncertainty Estimation Type=Aleatoric2026.03 | 76.9 | 6.7 | 0.722 | |
| PostNet EpistemicBackbone=ResNet18, Uncertainty Estimation Type=Epistemic2026.03 | 75.1 | 15.8 | 0.698 | |
| IGRUEBackbone=ResNet182026.03 | 54.8 | 15.7 | 0.027 |