OOD Detection on CIFAR-10 (IND) vs TinyImageNet (OOD)
94.18AUROCNGC
Evaluation Results
| Method | Links | |
|---|---|---|
| NGCSupervised detection=false, IND noise level=50% sym., OOD samples in training set=20k, OOD samples in test set=10k2021.08 | 94.18 | |
| MDSupervised detection=true, IND noise level=50% sym., OOD samples in training set=20k, OOD samples in test set=10k2021.08 | 77.55 | |
| SSDSupervised detection=false, IND noise level=50% sym., OOD samples in training set=20k, OOD samples in test set=10k2021.08 | 75.51 | |
| MSPSupervised detection=true, IND noise level=50% sym., OOD samples in training set=20k, OOD samples in test set=10k2021.08 | 70.12 | |
| RotSupervised detection=false, IND noise level=50% sym., OOD samples in training set=20k, OOD samples in test set=10k2021.08 | 68.87 | |
| ODINSupervised detection=true, IND noise level=50% sym., OOD samples in training set=20k, OOD samples in test set=10k2021.08 | 67.31 | |
| RotSupervised detection=true, IND noise level=50% sym., OOD samples in training set=20k, OOD samples in test set=10k2021.08 | 64.64 | |
| SSDSupervised detection=true, IND noise level=50% sym., OOD samples in training set=20k, OOD samples in test set=10k2021.08 | 60.52 |