Image Classification on ImageNet (Accuracy and Calibration Metrics)
0.817AccuracyDeep Ensembles
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Deep EnsemblesBackbone=ViT2023.12 | 0.817 | 0.135 | 0.217 | 0.855 | 0.603 | |
| LaplaceBackbone=ViT2023.12 | 0.81 | 0.108 | 0.221 | 0.831 | 0.706 | |
| ABNNBackbone=ViT2023.12 | 0.806 | 0.0432 | 0.217 | 0.854 | 0.551 | |
| LaplaceBackbone=ResNet-502023.12 | 0.804 | 0.443 | 0.139 | 0.759 | 0.828 | |
| Single ModelBackbone=ViT2023.12 | 0.8 | 0.052 | 0.195 | 0.841 | 0.585 | |
| ABNNBackbone=ResNet-502023.12 | 0.795 | 0.0965 | 0.178 | 0.82 | 0.652 | |
| Deep EnsemblesBackbone=ResNet-502023.12 | 0.792 | 0.233 | 0.196 | 0.834 | 0.621 | |
| Single ModelBackbone=ResNet-502023.12 | 0.778 | 0.121 | 0.18 | 0.809 | 0.686 | |
| MIMO (P = 1)Backbone=ResNet-502023.12 | 0.776 | 0.147 | 0.184 | 0.816 | 0.668 | |
| BatchEnsembleBackbone=ResNet-502023.12 | 0.759 | 0.035 | 0.202 | 0.816 | 0.665 |