OOD Detection on MNIST (ID) vs SVHN (OOD) (test)
0.9997AUCBNN-ARHT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BNN-ARHTIn-Distrib.=MNIST, Architecture=LeNet [22]2023.10 | 0.9997 | 0.9996 | |
| MC DropoutIn-Distrib.=MNIST, Architecture=LeNet [22]2023.10 | 0.9996 | 0.9996 | |
| DPNIn-Distrib.=MNIST, Architecture=LeNet [22]2023.10 | 0.9996 | 0.9996 | |
| Kendall and GalIn-Distrib.=MNIST, Architecture=LeNet [22]2023.10 | 0.996 | 0.9913 | |
| Deep EnsemblesIn-Distrib.=MNIST, Architecture=LeNet [22]2023.10 | 0.9921 | 0.9968 | |
| DetectronIn-Distrib.=MNIST, Architecture=LeNet [22]2023.10 | 0.7792 | 0.8375 | |
| EDLIn-Distrib.=MNIST, Architecture=LeNet [22]2023.10 | 0.6343 | 0.8509 |