Image Classification on CIFAR-10 LT λ=20 (test)
12.36Error RateMetaSAug with cross-entropy loss
Evaluation results
| Method | Links | |
|---|---|---|
| MetaSAug with cross-entropy lossBackbone=ResNet-322021.12 | 12.36 | |
| RISDABackbone=ResNet-322021.12 | 13.02 | |
| Meta-class-weight with cross-entropy lossBackbone=ResNet-322021.12 | 13.54 | |
| Class-balanced CEBackbone=ResNet-322021.12 | 15.44 | |
| Meta-weight netBackbone=ResNet-322021.12 | 15.55 | |
| Class-balanced fine-tuningBackbone=ResNet-322021.12 | 16.78 | |
| Cross-EntropyBackbone=ResNet-322021.12 | 17.56 | |
| L2RWBackbone=ResNet-322021.12 | 18.65 |