Image Classification on CIFAR-100 IR=10, NR=0.1 (test)
39.6Top-1 Error Rateλ^{A,W,S}
Evaluation results
| Method | Links | |
|---|---|---|
| λ^{A,W,S}Backbone=WRNet28-10, Configuration=per-point augmentation + loss reweighting + soft-labeling2021.03 | 39.6 | |
| λ^ABackbone=WRNet28-10, Configuration=per-point augmentation hyperparameters2021.03 | 47.4 | |
| λ^{A,W}Backbone=WRNet28-10, Configuration=per-point augmentation + loss reweighting2021.03 | 48 | |
| FAABackbone=WRNet28-102021.03 | 48.9 | |
| λ^{ASHA}Backbone=WRNet28-10, Configuration=shared augmentation hyperparameters2021.03 | 48.9 | |
| DADABackbone=WRNet28-102021.03 | 49.6 | |
| RAABackbone=WRNet28-102021.03 | 50.1 | |
| BaselineBackbone=WRNet28-102021.03 | 54.5 |