List your company
ResearchFormerly WizWand

Image Classification on CIFAR-10 Practical Heterogeneity

89.78AccuracypFedMe
79.27682.00384.7387.457Jul 8, 2026
Updated 3mo ago

Evaluation results

MethodLinks
pFedMeHeterogeneity Type=Practical, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.0789.78
FedAMPHeterogeneity Type=Practical, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.0789.14
FedKT-CSDHeterogeneity Type=Practical, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=1 round (synthetic pretraining) + local fine-tuning2026.0788.78
APPLEHeterogeneity Type=Practical, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.0788.57
Per-FedAvgHeterogeneity Type=Practical, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.0787.93
FedPHPHeterogeneity Type=Practical, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.0787.69
FedFomoHeterogeneity Type=Practical, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.0787.54
PartialFedHeterogeneity Type=Practical, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.0787.29
FedAvgHeterogeneity Type=Practical, Backbone=ResNet-18, Number of clients=20, Participation=Full, Communication rounds=502026.0779.68