OOD Detection on CIFAR-100 OOD (test)
98.6AUROCGeneralized (Neg)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Generalized (Neg)Backbone=ResNet18-SimCLR, ID Dataset=STL102026.03 | 98.6 | 4.7 | |
| Single (Neg)Backbone=ResNet18-SimCLR, ID Dataset=STL102026.03 | 93.3 | 59 | |
| MSP*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 91.05 | 33.27 | |
| Generalized (Neg, InfoNCE)Backbone=ResNet18-SimCLR, ID Dataset=STL102026.03 | 91 | 49.4 | |
| KNN*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 90.96 | 33.23 | |
| GEN*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 90.96 | 33.58 | |
| SCALE*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 90.86 | 34.17 | |
| ReAct*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 90.79 | 33.97 | |
| fDBD*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 90.31 | 36.28 | |
| KNNImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 89.73 | 37.64 | |
| fDBDImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 89.56 | 39.61 | |
| GENImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 87.21 | 58.75 | |
| MSPImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 87.19 | 53.08 | |
| ReActImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 85.93 | 67.4 | |
| COMBOODBackbone=ResNet18-SimCLR, ID Dataset=STL102026.03 | 85 | 49.3 | |
| SCALEImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 81.27 | 81.78 | |
| D-KNNBackbone=ResNet18-SimCLR, ID Dataset=STL102026.03 | 78.6 | 53.7 | |
| MSPBackbone=ResNet18-SimCLR, ID Dataset=STL102026.03 | 69.8 | 68.5 | |
| Energy (T=1)Backbone=ResNet18-SimCLR, ID Dataset=STL102026.03 | 69.5 | 68.4 | |
| CIDERBackbone=ResNet18-SimCLR, ID Dataset=STL102026.03 | 66.2 | 78.2 | |
| KNN (ℓ2)Backbone=ResNet18-SimCLR, ID Dataset=STL102026.03 | 54.2 | 87.5 | |
| MahalanobisBackbone=ResNet18-SimCLR, ID Dataset=STL102026.03 | 49.8 | 88.2 |