Out-of-Distribution Detection on CIFAR 10 (Near OOD)
98.1AUROCE3 normed-fork
Evaluation results
| Method | Links | ||
|---|---|---|---|
| E3 normed-forkBackbone=ResNet-50, Feature Dimension=2048-d2026.05 | 98.1 | 9 | |
| MahalanobisBackbone=ResNet-50, In-Distribution Dataset=CIFAR-102026.05 | 98.1 | 9.3 | |
| Score Network (Hk = 4dk)Hk=4dk2026.05 | 97.3 | 12.7 | |
| EncMin2L2026.05 | 96.3 | 19.5 | |
| AOE-JtMethod Category=Outlier Exposure, Training Strategy=joint training2026.05 | 95.64 | 17.44 | |
| AOE-AtMethod Category=Outlier Exposure, Training Strategy=alternating training2026.05 | 95.21 | 18.37 | |
| DOEMethod Category=Outlier Exposure2026.05 | 94.84 | 20.39 | |
| OCLMethod Category=Outlier Exposure2026.05 | 94.84 | 22.64 | |
| OEMethod Category=Outlier Exposure2026.05 | 94.82 | 19.84 | |
| DALMethod Category=Outlier Exposure2026.05 | 94.42 | 20.91 | |
| AdaNegExtra Resources=Yes2026.04 | 94.01 | 32.38 | |
| TTLExtra Resources=No2026.04 | 93.6 | 30.25 | |
| k-NNBackbone=ResNet-50, k=502026.05 | 93.6 | 39.5 | |
| NeglabelExtra Resources=Yes2026.04 | 92.96 | 35.32 | |
| T2FNormMethod Category=Training-time regularization2026.05 | 92.79 | 26.47 | |
| LogitNormMethod Category=Training-time regularization2026.05 | 92.33 | 29.34 | |
| k-NNBackbone=CLIP, k=502026.05 | 92.1 | 61.2 | |
| PSKDMethod Category=Training-time regularization2026.05 | 91.71 | 31.67 | |
| MSPBackbone=ResNet-50, In-Distribution Dataset=CIFAR-102026.05 | 91.5 | 53.4 | |
| MCMExtra Resources=No2026.04 | 91 | 35 | |
| UMAPMethod Category=Training-time regularization2026.05 | 91 | 33.01 | |
| AdaNDExtra Resources=No2026.04 | 90.97 | 35.06 | |
| E1 normed-forkBackbone=CLIP, Feature Dimension=512-d2026.05 | 90.7 | 55.1 | |
| UMMethod Category=Training-time regularization2026.05 | 90.6 | 33.12 | |
| OODDExtra Resources=No2026.04 | 89.19 | 48.61 | |
| Ensembles2026.05 | 88.89 | — | |
| fDBD2026.05 | 88.87 | — | |
| MixOEMethod Category=Outlier Exposure2026.05 | 88.73 | 51.45 | |
| EnergyBackbone=ResNet-50, In-Distribution Dataset=CIFAR-102026.05 | 88.5 | 58.3 | |
| KNN2026.05 | 88.07 | — | |
| MSPMethod Category=Post-hoc2026.05 | 88.03 | 48.17 | |
| HCM mixmixup=true2026.05 | 87.9 | — | |
| EnergyMethod Category=Post-hoc2026.05 | 87.58 | 61.34 | |
| Energy2026.05 | 87.52 | — | |
| MSP2026.05 | 86.73 | — | |
| k-NNBackbone=DINOv2, k=502026.05 | 86.7 | 74.6 | |
| NCI2026.05 | 86.49 | — | |
| ODIN2026.05 | 85.49 | — | |
| MC2026.05 | 85.21 | — | |
| MDS2026.05 | 84.91 | — | |
| ViM2026.05 | 84.2 | — | |
| HCMmixup=false2026.05 | 82.23 | — | |
| Score Network (Hk = 2dk)Hk=2dk2026.05 | 63.1 | 94.9 | |
| E2 raw-forkBackbone=DINOv2, Feature Dimension=768-d2026.05 | 56.9 | 95.7 |