Image Classification on CIFAR-10 Step (rho=50) (test)
17.9Top-1 Error RateLDAM-DRW
Evaluation results
| Method | Links | |
|---|---|---|
| LDAM-DRWBackbone=ResNet-32, RSG=true2021.06 | 17.9 | |
| LDAM-DRWBackbone=ResNet-32, RSG=false2021.06 | 18.67 | |
| M-DRWBackbone=ResNet-32, RSG=true2021.06 | 20.09 | |
| M-DRWBackbone=ResNet-32, RSG=false2021.06 | 21.05 | |
| Focal LossBackbone=ResNet-32, RSG=true2021.06 | 26.01 | |
| ERM-RSGBackbone=ResNet-32, Learning Protocol=ERM, Imbalance Ratio (rho)=502021.06 | 26.07 | |
| ERMBackbone=ResNet-32, RSG=true2021.06 | 26.07 | |
| A-EncoderBackbone=ResNet-32, Learning Protocol=ERM, Imbalance Ratio (rho)=502021.06 | 27.7 | |
| FTLBackbone=ResNet-32, Learning Protocol=ERM, Imbalance Ratio (rho)=502021.06 | 27.83 | |
| ImaginaryBackbone=ResNet-32, Learning Protocol=ERM, Imbalance Ratio (rho)=502021.06 | 28.23 | |
| Focal LossBackbone=ResNet-32, RSG=false2021.06 | 28.7 | |
| ERMBackbone=ResNet-32, RSG=false2021.06 | 28.88 |