Image Classification on CIFAR-100 semantic noise (test)
68.44AccuracyScanMix
Evaluation results
| Method | Links | |
|---|---|---|
| ScanMixBackbone=DenseNet, Noise ratio=34%2021.03 | 68.44 | |
| DivideMixBackbone=DenseNet, Noise ratio=34%2021.03 | 68.4 | |
| ScanMixBackbone=ResNet, Noise ratio=37%2021.03 | 67.36 | |
| ScanMixBackbone=VGG, Noise ratio=37%2021.03 | 67.34 | |
| DivideMixBackbone=VGG, Noise ratio=37%2021.03 | 66.84 | |
| DivideMixBackbone=ResNet, Noise ratio=37%2021.03 | 66.28 | |
| CE + ROGBackbone=DenseNet, Noise ratio=34%2021.03 | 61.14 | |
| Bootstrap + ROGBackbone=DenseNet, Noise ratio=34%2021.03 | 54.71 | |
| Backward + ROGBackbone=DenseNet, Noise ratio=34%2021.03 | 54.01 | |
| Forward + RoGBackbone=DenseNet, Noise ratio=34%2021.03 | 53.91 | |
| Bootstrap + ROGBackbone=VGG, Noise ratio=37%2021.03 | 53.76 | |
| CE + ROGBackbone=VGG, Noise ratio=37%2021.03 | 53.64 | |
| Forward + RoGBackbone=VGG, Noise ratio=37%2021.03 | 53.63 | |
| Backward + ROGBackbone=VGG, Noise ratio=37%2021.03 | 53.5 | |
| Forward + RoGBackbone=ResNet, Noise ratio=37%2021.03 | 53.36 | |
| Bootstrap + ROGBackbone=ResNet, Noise ratio=37%2021.03 | 53.3 | |
| CE + ROGBackbone=ResNet, Noise ratio=37%2021.03 | 53.09 | |
| Backward + ROGBackbone=ResNet, Noise ratio=37%2021.03 | 53.03 | |
| D2L + ROGBackbone=VGG, Noise ratio=37%2021.03 | 45.42 | |
| D2L + ROGBackbone=ResNet, Noise ratio=37%2021.03 | 39.92 | |
| D2L + ROGBackbone=DenseNet, Noise ratio=34%2021.03 | 31.67 |