Image Classification on CIFAR-10 (test) (Accuracy under Symmetric/Asymmetric Corruption)
95.91Accuracy (No Noise 0%)GJS
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| GJSBackbone=ResNet-342021.05 | 95.91 | 95.33 | 93.57 | 91.64 | 79.11 | 93.94 | 89.65 | |
| JSBackbone=ResNet-342021.05 | 95.89 | 94.52 | 93.01 | 89.64 | 76.06 | 92.18 | 87.99 | |
| CEBackbone=ResNet-342021.05 | 95.77 | 91.63 | 87.74 | 81.99 | 55.77 | 92.77 | 87.12 | |
| SCEBackbone=ResNet-342021.05 | 95.75 | 94.29 | 92.72 | 89.26 | 80.68 | 93.48 | 84.98 | |
| GCEBackbone=ResNet-342021.05 | 95.75 | 94.24 | 92.82 | 89.37 | 79.19 | 92.83 | 87 | |
| LSBackbone=ResNet-342021.05 | 95.64 | 93.51 | 89.9 | 83.96 | 67.35 | 92.94 | 88.1 | |
| NCE+RCEBackbone=ResNet-342021.05 | 95.36 | 94.27 | 92.03 | 87.3 | 77.89 | 93.87 | 86.83 | |
| BSBackbone=ResNet-342021.05 | 94.58 | 91.68 | 89.23 | 82.65 | 65.74 | 93.06 | 88.87 |