Image Classification on CIFAR-10 (test) (ACC, AUC, AsCC, AsMC)
85.64AccuracyMSCE
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MSCEBackbone=ResNet-182026.03 | 85.64 | 98.8 | 89.12 | 35.7 | |
| CDW-CEBackbone=ResNet-182026.03 | 84.11 | 98.57 | 87.87 | 34.63 | |
| CO2Backbone=ResNet-182026.03 | 83.97 | 98.55 | 87.77 | 34.73 | |
| Cross Entropy (CE)Backbone=ResNet-182026.03 | 83.24 | 98.57 | 87.23 | 34.84 | |
| Weighted CE (1:10)Backbone=ResNet-18, Weight Ratio=1:102026.03 | 73.69 | 95.9 | 79.79 | 33.55 |