Image Classification on CIFAR-100-C (Mean over Intensities 1-5)
49.33AccuracyRegular Ensemble
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Regular EnsembleBackbone=ResNet-20-FRN2022.12 | 49.33 | 2.067 | 0.115 | 53.3 | 39.7 | |
| Dropout (SE1)Backbone=ResNet-20-FRN2022.12 | 48.35 | 2.115 | 0.089 | 61.1 | 32 | |
| Multi SWABackbone=ResNet-20-FRN2022.12 | 47.48 | 2.153 | 0.112 | 53.4 | 39.9 | |
| HMC 3-chainsBackbone=ResNet-20-FRN, Configuration=3-chains2022.12 | 42.1 | 2.61 | 0.113 | 100 | 0 |