Image Classification on mini-ImageNet-Blue (0% noise)
73.4AccuracyNCR
Evaluation results
| Method | Links | |
|---|---|---|
| NCRBackbone=ResNet-18, Training Protocol=NCR2022.02 | 73.4 | |
| NCR + MixupBackbone=ResNet-18, Training Protocol=NCR, Data Augmentation=Mixup2022.02 | 73.1 | |
| Label smoothingBackbone=ResNet-18, Training Protocol=Baseline2022.02 | 73 | |
| BootstrapBackbone=ResNet-18, Training Protocol=Baseline2022.02 | 72.8 | |
| StandardBackbone=ResNet-18, Training Protocol=Baseline2022.02 | 72.7 | |
| StandardBackbone=ResNet-18, Training Protocol=Oracle: Clean Subset2022.02 | 72.7 | |
| MixupBackbone=ResNet-18, Training Protocol=Baseline, Data Augmentation=Mixup2022.02 | 72.6 | |
| MixupBackbone=ResNet-18, Training Protocol=Oracle: Clean Subset, Data Augmentation=Mixup2022.02 | 72.6 | |
| Label smoothing + MixupBackbone=ResNet-18, Training Protocol=Baseline, Data Augmentation=Mixup2022.02 | 72.5 | |
| Bootstrap + MixupBackbone=ResNet-18, Training Protocol=Baseline, Data Augmentation=Mixup2022.02 | 71.7 |