Model Calibration on CIFAR-10
0.76ECEBalCAL
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| BalCALBackbone=WRN-28-102025.04 | 0.76 | — | — | — | — | 92.23 | 0.82 | — | |
| Knowledge-Transferring-based Temperature ScalingIF=102023.04 | 1 | — | — | — | — | — | — | — | |
| VTSTBackbone=WRN-28-102025.04 | 1.52 | — | — | — | — | 91.09 | — | — | |
| ETSIF=102023.04 | 1.64 | — | — | — | — | — | — | — | |
| FLSDBackbone=WRN-28-102025.04 | 1.72 | — | — | — | — | 92.42 | 1.68 | — | |
| FLBackbone=WRN-28-102025.04 | 1.99 | — | — | — | — | 92.11 | 1.97 | — | |
| GPCIF=102023.04 | 2.01 | — | — | — | — | — | — | — | |
| TS-IRIF=102023.04 | 2.03 | — | — | — | — | — | — | — | |
| MITBackbone=WRN-28-102025.04 | 2.18 | — | — | — | — | 94.27 | 2.12 | — | |
| TSTBackbone=WRN-28-102025.04 | 2.18 | — | — | — | — | 92.59 | — | — | |
| TSIF=102023.04 | 2.23 | — | — | — | — | — | — | — | |
| IRIF=102023.04 | 2.29 | — | — | — | — | — | — | — | |
| IROVAIF=102023.04 | 2.42 | — | — | — | — | — | — | — | |
| SBCIF=102023.04 | 2.49 | — | — | — | — | — | — | — | |
| MixupBackbone=WRN-28-102025.04 | 2.56 | — | — | — | — | 94.71 | 3.08 | — | |
| LSBackbone=WRN-28-102025.04 | 2.79 | — | — | — | — | 92.12 | 4.26 | — | |
| MMCEBackbone=WRN-28-102025.04 | 3.6 | — | — | — | — | 90.04 | 3.43 | — | |
| Knowledge-Transferring-based Temperature ScalingIF=502023.04 | 3.99 | — | — | — | — | — | — | — | |
| PLPBackbone=WRN-28-102025.04 | 4.89 | — | — | — | — | 92.69 | 4.85 | — | |
| ACLSBackbone=WRN-28-102025.04 | 5.26 | — | — | — | — | 93.12 | 5.26 | — | |
| CPCBackbone=WRN-28-102025.04 | 6.28 | — | — | — | — | 91.56 | 6.25 | — | |
| VanillaBackbone=WRN-28-102025.04 | 6.38 | — | — | — | — | 91.34 | 6.38 | — | |
| SBCIF=502023.04 | 7.55 | — | — | — | — | — | — | — | |
| TSIF=502023.04 | 7.65 | — | — | — | — | — | — | — | |
| GPCIF=502023.04 | 7.78 | — | — | — | — | — | — | — | |
| ETSIF=502023.04 | 8.04 | — | — | — | — | — | — | — | |
| TS-IRIF=502023.04 | 8.22 | — | — | — | — | — | — | — | |
| BaseIF=102023.04 | 8.39 | — | — | — | — | — | — | — | |
| IROVAIF=502023.04 | 9.45 | — | — | — | — | — | — | — | |
| IRIF=502023.04 | 9.75 | — | — | — | — | — | — | — | |
| Knowledge-Transferring-based Temperature ScalingIF=1002023.04 | 9.84 | — | — | — | — | — | — | — | |
| TS-IRIF=1002023.04 | 11.64 | — | — | — | — | — | — | — | |
| GPCIF=1002023.04 | 11.65 | — | — | — | — | — | — | — | |
| SBCIF=1002023.04 | 12.13 | — | — | — | — | — | — | — | |
| ETSIF=1002023.04 | 12.16 | — | — | — | — | — | — | — | |
| TSIF=1002023.04 | 12.24 | — | — | — | — | — | — | — | |
| IRIF=1002023.04 | 12.36 | — | — | — | — | — | — | — | |
| IROVAIF=1002023.04 | 13.36 | — | — | — | — | — | — | — | |
| BaseIF=502023.04 | 17.36 | — | — | — | — | — | — | — | |
| BaseIF=1002023.04 | 21.79 | — | — | — | — | — | — | — | |
| CEModel=Wide-ResNet2026.04 | 55 | — | — | — | — | — | — | 3.44 | |
| AFLModel=Wide-ResNet2026.04 | 56 | — | — | — | — | — | — | 0.61 | |
| FLSDModel=Wide-ResNet2026.04 | 59 | — | — | — | — | — | — | 1.98 | |
| CEModel=ResNet-502026.04 | 66 | — | — | — | — | — | — | 4.25 | |
| GCEModel=ResNet-502026.04 | 66 | — | — | — | — | — | — | 3.25 | |
| FLSDModel=ResNet-1102026.04 | 66 | — | — | — | — | — | — | 1.64 | |
| GCEModel=ResNet-1102026.04 | 66 | — | — | — | — | — | — | 3.69 | |
| FLSDModel=ResNet-502026.04 | 67 | — | — | — | — | — | — | 1.53 | |
| AFLModel=ResNet-502026.04 | 67 | — | — | — | — | — | — | 0.83 | |
| DFLModel=Wide-ResNet2026.04 | 67 | — | — | — | — | — | — | 1.01 | |
| DFLModel=ResNet-502026.04 | 68 | — | — | — | — | — | — | 1.06 | |
| Brier LossModel=Wide-ResNet2026.04 | 68 | — | — | — | — | — | — | 1.23 | |
| MMCEModel=Wide-ResNet2026.04 | 68 | — | — | — | — | — | — | 3.6 | |
| GCEModel=Wide-ResNet2026.04 | 68 | — | — | — | — | — | — | 2.66 | |
| Brier LossModel=ResNet-502026.04 | 70 | — | — | — | — | — | — | 2 | |
| DFLModel=DenseNet-1212026.04 | 70 | — | — | — | — | — | — | 0.95 | |
| Brier LossModel=ResNet-1102026.04 | 71 | — | — | — | — | — | — | 2.24 | |
| GCEModel=DenseNet-1212026.04 | 71 | — | — | — | — | — | — | 3.18 | |
| MMCEModel=ResNet-1102026.04 | 73 | — | — | — | — | — | — | 4.89 | |
| AFLModel=DenseNet-1212026.04 | 73 | — | — | — | — | — | — | 0.1 | |
| DFLModel=ResNet-1102026.04 | 75 | — | — | — | — | — | — | 1.57 | |
| FLSDModel=DenseNet-1212026.04 | 75 | — | — | — | — | — | — | 1.42 | |
| AFLModel=ResNet-1102026.04 | 77 | — | — | — | — | — | — | 0.84 | |
| MMCEModel=ResNet-502026.04 | 78 | — | — | — | — | — | — | 4.61 | |
| MMCEModel=DenseNet-1212026.04 | 79 | — | — | — | — | — | — | 4.98 | |
| CEModel=DenseNet-1212026.04 | 81 | — | — | — | — | — | — | 4.63 | |
| CEModel=ResNet-1102026.04 | 83 | — | — | — | — | — | — | 4.71 | |
| Brier LossModel=DenseNet-1212026.04 | 83 | — | — | — | — | — | — | 1.72 | |
| Adv TrainingBackbone=AllConvNet2019.12 | — | 11.1 | — | — | — | — | — | — | |
| Adv TrainingBackbone=DenseNet2019.12 | — | 16.2 | — | — | — | — | — | — | |
| Adv TrainingBackbone=WideResNet2019.12 | — | 10.7 | — | — | — | — | — | — | |
| Adv TrainingBackbone=ResNeXt2019.12 | — | 5.8 | — | — | — | — | — | — | |
| Adv TrainingBackbone=Mean2019.12 | — | 11 | — | — | — | — | — | — | |
| AugMix2021.12 | — | — | — | — | 9.4 | — | — | — | |
| AUGMIXBackbone=AllConvNet2019.12 | — | 2.2 | — | — | — | — | — | — | |
| AUGMIXBackbone=DenseNet2019.12 | — | 5 | — | — | — | — | — | — | |
| AUGMIXBackbone=WideResNet2019.12 | — | 4.2 | — | — | — | — | — | — | |
| AUGMIXBackbone=ResNeXt2019.12 | — | 3 | — | — | — | — | — | — | |
| AUGMIXBackbone=Mean2019.12 | — | 3.6 | — | — | — | — | — | — | |
| Auto Augment2021.12 | — | — | — | — | 14.8 | — | — | — | |
| AutoAugment*Backbone=AllConvNet2019.12 | — | 4.2 | — | — | — | — | — | — | |
| AutoAugment*Backbone=DenseNet2019.12 | — | 6 | — | — | — | — | — | — | |
| AutoAugment*Backbone=WideResNet2019.12 | — | 4.7 | — | — | — | — | — | — | |
| AutoAugment*Backbone=ResNeXt2019.12 | — | 3.3 | — | — | — | — | — | — | |
| AutoAugment*Backbone=Mean2019.12 | — | 4.6 | — | — | — | — | — | — | |
| Baseline2021.12 | — | — | — | — | 22.7 | — | — | — | |
| CutMixBackbone=AllConvNet2019.12 | — | 3.1 | — | — | — | — | — | — | |
| CutMixBackbone=DenseNet2019.12 | — | 5.4 | — | — | — | — | — | — | |
| CutMixBackbone=WideResNet2019.12 | — | 5 | — | — | — | — | — | — | |
| CutMixBackbone=ResNeXt2019.12 | — | 3.5 | — | — | — | — | — | — | |
| CutMixBackbone=Mean2019.12 | — | 4.2 | — | — | — | — | — | — | |
| CutMix2021.12 | — | — | — | — | 18.6 | — | — | — | |
| CutoutBackbone=AllConvNet2019.12 | — | 4 | — | — | — | — | — | — | |
| CutoutBackbone=DenseNet2019.12 | — | 6.4 | — | — | — | — | — | — | |
| CutoutBackbone=WideResNet2019.12 | — | 3.8 | — | — | — | — | — | — | |
| CutoutBackbone=ResNeXt2019.12 | — | 4.4 | — | — | — | — | — | — | |
| CutoutBackbone=Mean2019.12 | — | 4.7 | — | — | — | — | — | — | |
| Cutout2021.12 | — | — | — | — | 17.8 | — | — | — | |
| MixupBackbone=AllConvNet2019.12 | — | 12.6 | — | — | — | — | — | — | |
| MixupBackbone=DenseNet2019.12 | — | 15.6 | — | — | — | — | — | — |