OOD Detection on CIFAR10 (ID) vs Fashion-MNIST (OOD) (test)
77.78AUCBNN-ARHT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BNN-ARHTIn-Distrib.=CIFAR10, Architecture=LeNet [22]2023.10 | 77.78 | 79.06 | |
| Kendall and GalIn-Distrib.=CIFAR10, Architecture=LeNet [22]2023.10 | 77.41 | 77 | |
| DetectronIn-Distrib.=CIFAR10, Architecture=LeNet [22]2023.10 | 76.46 | 71.63 | |
| MC DropoutIn-Distrib.=CIFAR10, Architecture=LeNet [22]2023.10 | 76.23 | 74.21 | |
| Deep EnsemblesIn-Distrib.=CIFAR10, Architecture=LeNet [22]2023.10 | 71.25 | 75.32 | |
| EDLIn-Distrib.=CIFAR10, Architecture=LeNet [22]2023.10 | 67.81 | 71.81 | |
| DPNIn-Distrib.=CIFAR10, Architecture=LeNet [22]2023.10 | 57.54 | 68.29 |