Federated Image Classification on SVHN (test) (Efficiency and Convergence Metrics)
1RoundsProposed
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ProposedOverlap=0.0, Dirichlet α=0.7, FL Alg.=MOON2024.10 | 1 | 6.9 | 0.08 | — | — | |
| ProposedOverlap=0.2, Dirichlet α=0.3, FL Alg.=FedProx2024.10 | 1 | 6.9 | 0.08 | — | — | |
| Previous (Baseline)Overlap=0.2, Dirichlet α=0.3, FL Alg.=FedProx2024.10 | 4 | 27.5 | 0.31 | 4 | 8 | |
| Previous (Baseline)Overlap=0.0, Dirichlet α=0.7, FL Alg.=MOON2024.10 | 5 | 34.4 | 0.39 | 5 | 11.5 | |
| Average (Baseline)Overlap=0.2, Dirichlet α=0.3, FL Alg.=FedProx2024.10 | 10 | 68.9 | 0.78 | 10 | 115 | |
| Average (Baseline)Overlap=0.0, Dirichlet α=0.7, FL Alg.=MOON2024.10 | 14 | 96.4 | 1.09 | 14 | 116.1 | |
| Continuous (Baseline)Overlap=0.0, Dirichlet α=0.7, FL Alg.=MOON2024.10 | 52 | 358.1 | 4.04 | 52 | 395.1 | |
| Continuous (Baseline)Overlap=0.2, Dirichlet α=0.3, FL Alg.=FedProx2024.10 | 70 | 482.1 | 5.44 | 70 | 517.6 |