Image Classification on CIFAR-10 Dir 0.8
53.16Top-1 AccuracyFedAvg
Evaluation results
| Method | Links | |
|---|---|---|
| FedAvgN (Number of clients)=10, Communication Rounds (R)=30, Batch size=64, Local epochs (E)=5, Optimiser=SGD, mom. 0.9, Learning rate (η)=0.012026.05 | 53.16 | |
| FedRepN (Number of clients)=10, Communication Rounds (R)=30, Batch size=64, Local epochs (Eh/Ee)=2/2, Optimiser=SGD, mom. 0.9, Learning rate (η)=0.012026.05 | 42.21 | |
| FibFL++N (Number of clients)=10, Communication Rounds (R)=30, Batch size=64, Head epochs (Eh)=1, Extractor epochs (Ee)=20, Optimiser=Adam (persistent), Learning rate (η)=0.012026.05 | 40.31 | |
| RDFLN (Number of clients)=10, Communication Rounds (R)=30, Batch size=64, Local epochs (E)=5, Optimiser=SGD, mom. 0.9, Learning rate (η)=0.01, γ (blend weight)=0.52026.05 | 30.55 | |
| FibFLN (Number of clients)=10, Communication Rounds (R)=30, Batch size=64, Head epochs (Eh)=1, Extractor epochs (Ee)=20, Optimiser=Adam (persistent), Learning rate (η)=0.012026.05 | 26.42 | |
| FibFL+N (Number of clients)=10, Communication Rounds (R)=30, Batch size=64, Head epochs (Eh)=1, Extractor epochs (Ee)=20, Optimiser=Adam (persistent), Learning rate (η)=0.012026.05 | 25.91 |