OOD Detection on CIFAR-10 (ID) vs TEXTURES (OOD)
0.96FPR@95Medix
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MedixTraining Data=Pin and Pwild2025.10 | 0.96 | 99.66 | |
| WOODSTraining Data=Pin and Pwild2025.10 | 6.21 | 98.13 | |
| OETraining Data=Pin and Pwild2025.10 | 7.75 | 98.56 | |
| KNN+Training Data=Pin only2025.10 | 9.65 | 98.37 | |
| Energy (w/ OE)Training Data=Pin and Pwild2025.10 | 10.51 | 97.1 | |
| MahalanobisTraining Data=Pin only2025.10 | 15 | 97.33 | |
| CSITraining Data=Pin only2025.10 | 20.45 | 95.93 | |
| ASHTraining Data=Pin only2025.10 | 24.34 | 95.09 | |
| ODINTraining Data=Pin only2025.10 | 49.12 | 84.97 | |
| KNNTraining Data=Pin only2025.10 | 50.9 | 89.14 | |
| EnergyTraining Data=Pin only2025.10 | 52.79 | 85.22 | |
| DICETraining Data=Pin only2025.10 | 53.62 | 82.2 | |
| ReActTraining Data=Pin only2025.10 | 53.63 | 86.59 | |
| MSPTraining Data=Pin only2025.10 | 59.28 | 88.5 |