Image Classification on CIFAR-10 (test) (Accuracy, ECE, SCE, Brier score)
93.3AccuracyDeep Ensemble
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Deep EnsembleBackbone=EfficientNet-B0, Ensemble size (M)=42022.05 | 93.3 | 0.0307 | 0.0067 | 0.1008 | |
| Deep EnsembleBackbone=EfficientNet-B0, Ensemble size (M)=22022.05 | 92.67 | 0.0373 | 0.008 | 0.1146 | |
| ASFLBase Model=ResNet-502026.02 | 92.52 | — | — | — | |
| FiLM-EnsembleBackbone=EfficientNet-B0, Ensemble size (M)=42022.05 | 91.73 | 0.0163 | 0.0044 | 0.1222 | |
| FiLM-EnsembleBackbone=EfficientNet-B0, Ensemble size (M)=22022.05 | 91.62 | 0.0336 | 0.0073 | 0.1291 | |
| SFLBase Model=ResNet-502026.02 | 91.07 | — | — | — | |
| ACC-SFLBase Model=ResNet-502026.02 | 90.94 | — | — | — | |
| FedAvgBase Model=ResNet-502026.02 | 90.91 | — | — | — | |
| MC-DropoutBackbone=EfficientNet-B0, Ensemble size (M)=22022.05 | 90.81 | 0.0499 | 0.0107 | 0.1478 | |
| MC-DropoutBackbone=EfficientNet-B0, Ensemble size (M)=42022.05 | 90.81 | 0.0497 | 0.0107 | 0.1474 | |
| SingleBackbone=EfficientNet-B0, Ensemble size (M)=12022.05 | 90.8 | 0.0496 | 0.0106 | 0.147 | |
| FedAvgBase Model=VGG-192026.02 | 89.69 | — | — | — | |
| EPSLBase Model=ResNet-502026.02 | 89.47 | — | — | — | |
| ASFLBase Model=VGG-192026.02 | 89.2 | — | — | — | |
| SLBase Model=ResNet-502026.02 | 88.81 | — | — | — | |
| SFLBase Model=VGG-192026.02 | 88.77 | — | — | — | |
| ACC-SFLBase Model=VGG-192026.02 | 88.02 | — | — | — | |
| EPSLBase Model=VGG-192026.02 | 87.59 | — | — | — | |
| SLBase Model=VGG-192026.02 | 83.41 | — | — | — |