Classification on PH2 (test)
0.96AUCG-CBM ResNet-50
Evaluation results
| Method | Links | ||
|---|---|---|---|
| G-CBM ResNet-50Backbone=ResNet-50, Concept Bottleneck=Interpretable concept-graph2026.07 | 0.96 | 92.5 | |
| CNN DenseNet-201Backbone=DenseNet-201, Concept Bottleneck=None2026.07 | 0.945 | 86.4 | |
| CNN MobileNet-V2Backbone=MobileNet-V2, Concept Bottleneck=None2026.07 | 0.943 | 90.8 | |
| G-CBM DenseNet-201Backbone=DenseNet-201, Concept Bottleneck=Interpretable concept-graph2026.07 | 0.937 | 90.8 | |
| G-CBM MobileNet-V2Backbone=MobileNet-V2, Concept Bottleneck=Interpretable concept-graph2026.07 | 0.913 | 88.9 | |
| CNN ResNet-50Backbone=ResNet-50, Concept Bottleneck=None2026.07 | 0.903 | 90.1 | |
| MLP-CBM ResNet-50Backbone=ResNet-50, Concept Bottleneck=MLP2026.07 | 0.897 | 85.4 | |
| Linear-CBM ResNet-50Backbone=ResNet-50, Concept Bottleneck=Linear2026.07 | 0.887 | 86.6 |