Prostate cancer grading on SICAPv2 source-imbalanced KD (test)
0.9205OAResnet50 (Exp2)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Resnet50 (Exp2)Method Type=Individually trained model, Backbone=ResNet-502025.05 | 0.9205 | 0.8988 | 0.9193 | 0.11 | |
| Resnet50 (Exp1)Method Type=Individually trained model, Backbone=ResNet-502025.05 | 0.9153 | 0.8948 | 0.9147 | 0.1098 | |
| UMKDMethod Type=KD, Expert Backbone=ResNet-50, Student Backbone=ResNet-182025.05 | 0.9102 | 0.9023 | 0.9094 | 0.1294 | |
| Resnet18 (Stu)Method Type=Individually trained model, Backbone=ResNet-182025.05 | 0.8958 | 0.8906 | 0.8949 | 0.1463 | |
| KDMethod Type=KD, Expert Backbone=ResNet-50, Student Backbone=ResNet-182025.05 | 0.8906 | 0.8839 | 0.8897 | 0.1624 | |
| RKDMethod Type=Feature-based KD, Expert Backbone=ResNet-50, Student Backbone=ResNet-182025.05 | 0.8854 | 0.8824 | 0.8844 | 0.169 | |
| SDDMethod Type=KD, Expert Backbone=ResNet-50, Student Backbone=ResNet-182025.05 | 0.8782 | 0.8666 | 0.8767 | 0.1594 | |
| DKDMethod Type=KD, Expert Backbone=ResNet-50, Student Backbone=ResNet-182025.05 | 0.8691 | 0.8501 | 0.8668 | 0.1739 | |
| FitNetMethod Type=Feature-based KD, Expert Backbone=ResNet-50, Student Backbone=ResNet-182025.05 | 0.7878 | 0.7841 | 0.7842 | 0.3136 |