Image Classification on CIFAR-10 Pathological Heterogeneity
92.82AccuracyFedKT-CSD
Evaluation Results
| Method | Links | |
|---|---|---|
| FedKT-CSDHeterogeneity Type=Pathological, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=1 round (synthetic pretraining) + local fine-tuning2026.07 | 92.82 | |
| pFedMeHeterogeneity Type=Pathological, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.07 | 91.21 | |
| FedFomoHeterogeneity Type=Pathological, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.07 | 90.78 | |
| Per-FedAvgHeterogeneity Type=Pathological, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.07 | 90.12 | |
| APPLEHeterogeneity Type=Pathological, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.07 | 90.02 | |
| FedAMPHeterogeneity Type=Pathological, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.07 | 89.85 | |
| FedPHPHeterogeneity Type=Pathological, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.07 | 89.12 | |
| PartialFedHeterogeneity Type=Pathological, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.07 | 88.95 | |
| FedAvgHeterogeneity Type=Pathological, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.07 | 86.05 |