Image Classification on CIFAR-10 Flip2 noise 1.0 (test)
90.93Top-1 AccMW-net
Evaluation Results
| Method | Links | |
|---|---|---|
| MW-netNoise level p=0.2, Backbone=ResNet-322022.11 | 90.93 | |
| CrossEntropyNoise level p=0.2, Backbone=ResNet-322022.11 | 90.71 | |
| Advisor NetworkNoise level p=0.2, Backbone=ResNet-322022.11 | 90.66 | |
| Advisor NetworkNoise level p=0.4, Backbone=ResNet-322022.11 | 89.72 | |
| MW-netNoise level p=0.4, Backbone=ResNet-322022.11 | 88.83 | |
| CrossEntropyNoise level p=0.4, Backbone=ResNet-322022.11 | 87.83 | |
| Advisor NetworkNoise level p=0.6, Backbone=ResNet-322022.11 | 87.75 | |
| MW-netNoise level p=0.6, Backbone=ResNet-322022.11 | 86.85 | |
| CrossEntropyNoise level p=0.6, Backbone=ResNet-322022.11 | 75.83 | |
| Advisor NetworkNoise level p=0.8, Backbone=ResNet-322022.11 | 73.83 | |
| MW-netNoise level p=0.8, Backbone=ResNet-322022.11 | 27.49 | |
| CrossEntropyNoise level p=0.8, Backbone=ResNet-322022.11 | 11.86 |