Misclassification Detection on CIFAR-100 (test)
88.2AUROCD_alpha
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| D_alphaProtocol=PBB, Classifier Accuracy=78%2021.06 | 88.2 | — | — | — | — | — | — | 35.7 | — | |
| D_betaProtocol=PBB, Classifier Accuracy=78%2021.06 | 87.4 | — | — | — | — | — | — | 36.7 | — | |
| ODINProtocol=PBB, Classifier Accuracy=78%2021.06 | 87.1 | — | — | — | — | — | — | 40.7 | — | |
| Softmax BaselineArchitecture=40-4 wide residual network, Nonlinearity=GELU, Optimizer=SGD with restarts2016.10 | 87 | 50 | 96 | 79 | 62 | 21 | 66 | — | — | |
| D_alphaProtocol=TBB, Classifier Accuracy=78%2021.06 | 87 | — | — | — | — | — | — | 40.6 | — | |
| ODINProtocol=TBB, Classifier Accuracy=78%2021.06 | 86.9 | — | — | — | — | — | — | 40.5 | — | |
| SRProtocol=TBB, Classifier Accuracy=78%2021.06 | 86.9 | — | — | — | — | — | — | 40.5 | — | |
| D_betaProtocol=TBB, Classifier Accuracy=78%2021.06 | 84.2 | — | — | — | — | — | — | 40.6 | — | |
| MHLNBProtocol=TBB, Classifier Accuracy=78%2021.06 | 82.6 | — | — | — | — | — | — | 66.7 | — | |
| ENERGYProtocol=PBB, Classifier Accuracy=78%2021.06 | 78.7 | — | — | — | — | — | — | 65.4 | — | |
| MHLNBProtocol=PBB, Classifier Accuracy=78%2021.06 | 50 | — | — | — | — | — | — | 94 | — | |
| DAEDL2025.10 | — | — | — | — | — | — | — | — | 86 | |
| Dropout2025.10 | — | — | — | — | — | — | — | — | 92 | |
| EDL2025.10 | — | — | — | — | — | — | — | — | 91.28 | |
| F-EDL2025.10 | — | — | — | — | — | — | — | — | 94.01 | |
| I-EDL2025.10 | — | — | — | — | — | — | — | — | 92.84 | |
| R-EDL2025.10 | — | — | — | — | — | — | — | — | 92.69 |