Image Classification on CIFAR-100 Dirichlet alpha=0.6 (test)
45.99Test AccuracySpecGradFilter
Evaluation results
| Method | Links | |
|---|---|---|
| SpecGradFilterBackbone=ResNet-20, Number of clients=100, Client participation ratio=0.1, Communication rounds=300, Dirichlet distribution parameter alpha=0.6, spectral filter ratio r=0.052026.07 | 45.99 | |
| FedDynBackbone=ResNet-20, Number of clients=100, Client participation ratio=0.1, Communication rounds=300, Dirichlet distribution parameter alpha=0.62026.07 | 42.78 | |
| SCAFFOLDBackbone=ResNet-20, Number of clients=100, Client participation ratio=0.1, Communication rounds=300, Dirichlet distribution parameter alpha=0.62026.07 | 40.04 | |
| FedCMBackbone=ResNet-20, Number of clients=100, Client participation ratio=0.1, Communication rounds=300, Dirichlet distribution parameter alpha=0.62026.07 | 38.25 | |
| FedProxBackbone=ResNet-20, Number of clients=100, Client participation ratio=0.1, Communication rounds=300, Dirichlet distribution parameter alpha=0.62026.07 | 36.6 | |
| FedLWSBackbone=ResNet-20, Number of clients=100, Client participation ratio=0.1, Communication rounds=300, Dirichlet distribution parameter alpha=0.62026.07 | 36.33 | |
| FedDiscoBackbone=ResNet-20, Number of clients=100, Client participation ratio=0.1, Communication rounds=300, Dirichlet distribution parameter alpha=0.62026.07 | 36.27 | |
| FedAWABackbone=ResNet-20, Number of clients=100, Client participation ratio=0.1, Communication rounds=300, Dirichlet distribution parameter alpha=0.62026.07 | 36.1 | |
| FedSAMBackbone=ResNet-20, Number of clients=100, Client participation ratio=0.1, Communication rounds=300, Dirichlet distribution parameter alpha=0.62026.07 | 35.59 | |
| FedAvgBackbone=ResNet-20, Number of clients=100, Client participation ratio=0.1, Communication rounds=300, Dirichlet distribution parameter alpha=0.62026.07 | 35.37 |