OOD Detection on CIFAR-10 (AUROC, FPR95, ID-ACC)
12.89FPR@95SCT
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SCTShots=16-shot2024.11 | 12.89 | 96.1 | 93.1 | |
| OELearning Setup=Training methods with outliers2025.11 | 13.13 | 96 | — | |
| LogitNormLearning Setup=Training methods without outliers2025.11 | 13.81 | 96.74 | — | |
| LoCoOpShots=16-shot2024.11 | 15.3 | 95.37 | 93 | |
| MCMShots=16-shot2024.11 | 16.36 | 95.68 | 90.1 | |
| CRAFTLearning Setup=Training methods without outliers2025.11 | 19.4 | 93.94 | — | |
| UDGLearning Setup=Training methods with outliers2025.11 | 20.35 | 94.06 | — | |
| CIDERLearning Setup=Training methods without outliers2025.11 | 20.72 | 94.71 | — | |
| RankOODLearning Setup=Training methods without outliers2025.11 | 20.96 | 93.19 | — | |
| G-ODINLearning Setup=Training methods without outliers2025.11 | 21.45 | 95.51 | — | |
| ConfBranchLearning Setup=Training methods without outliers2025.11 | 21.48 | 92.85 | — | |
| KNNLearning Setup=Post-hoc inference methods2025.11 | 24.27 | 92.96 | — | |
| VIMLearning Setup=Post-hoc inference methods2025.11 | 25.05 | 93.48 | — | |
| RMDSLearning Setup=Post-hoc inference methods2025.11 | 25.35 | 92.2 | — | |
| MSPLearning Setup=Post-hoc inference methods2025.11 | 31.72 | 90.73 | — | |
| MCDLearning Setup=Training methods with outliers2025.11 | 32.03 | 91 | — | |
| TempScaleLearning Setup=Post-hoc inference methods2025.11 | 33.48 | 90.97 | — | |
| MixOELearning Setup=Training methods with outliers2025.11 | 33.84 | 91.93 | — | |
| GENLearning Setup=Post-hoc inference methods2025.11 | 34.73 | 91.35 | — | |
| ExCeLLearning Setup=Post-hoc inference methods2025.11 | 40.03 | 91.69 | — | |
| RASBackbone=ResNet182026.03 | 40.16 | 90.24 | — | |
| MLSLearning Setup=Post-hoc inference methods2025.11 | 41.68 | 91.1 | — | |
| EBOLearning Setup=Post-hoc inference methods2025.11 | 41.69 | 91.21 | — | |
| ReActLearning Setup=Post-hoc inference methods2025.11 | 44.9 | 90.42 | — | |
| EBOBackbone=ResNet182026.03 | 48.24 | 90 | — | |
| ReActBackbone=ResNet182026.03 | 51.12 | 89.31 | — | |
| ASH-PBackbone=ResNet182026.03 | 52.55 | 89.47 | — | |
| DICEBackbone=ResNet182026.03 | 57.85 | 82.27 | — | |
| SCALEBackbone=ResNet182026.03 | 71.77 | 85.18 | — | |
| ASH-SBackbone=ResNet182026.03 | 77.47 | 82.9 | — | |
| ASHLearning Setup=Post-hoc inference methods2025.11 | 79.03 | 78.49 | — | |
| ASH-BBackbone=ResNet182026.03 | 81.48 | 77.5 | — |