Medical Image Classification on Br35H
98.33Clean AccuracypFedDef
Evaluation results
| Method | Links | ||
|---|---|---|---|
| pFedDefStep Size (alpha)=0.003, Attack Type=L-infinity PGD, Number of steps (K)=7, Perturbation budget (epsilon)=0.015, Backbone=ResNet-182025.11 | 98.33 | 49.5 | |
| MedFedPureStep Size (alpha)=0.003, Attack Type=L-infinity PGD, Number of steps (K)=7, Perturbation budget (epsilon)=0.015, Backbone=ResNet-182025.11 | 97.67 | 87.33 |