Federated Learning Convergence Analysis on CIFAR10 Mild non-i.i.d. partition
5.9Cycles to 51% AccSSD-FL
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SSD-FLtau_r=5, Heterogeneous ML optimizers=true2026.06 | 5.9 | 8.57 | 11.63 | 15.88 | |
| pDFLtau_r=5, Heterogeneous ML optimizers=true2026.06 | 5.99 | 8.66 | 11.79 | 16.16 | |
| sDFLtau_r=5, Heterogeneous ML optimizers=true2026.06 | 6.2 | 8.9 | 12.11 | 16.61 | |
| cSTCtau_r=5, Heterogeneous ML optimizers=true2026.06 | 6.25 | 9.89 | 13.85 | 17.92 | |
| SSD-FLtau_r=3, Heterogeneous ML optimizers=true2026.06 | 6.41 | 9.26 | 12.87 | 17.73 | |
| cSTCtau_r=3, Heterogeneous ML optimizers=true2026.06 | 6.55 | 10.36 | 14.52 | 19.37 | |
| pDFLtau_r=3, Heterogeneous ML optimizers=true2026.06 | 6.61 | 9.57 | 13.62 | 18.99 | |
| STCtau_r=5, Heterogeneous ML optimizers=true2026.06 | 6.62 | 10.87 | 15.07 | 19.86 | |
| sDFLtau_r=3, Heterogeneous ML optimizers=true2026.06 | 6.68 | 9.71 | 13.79 | 18.64 | |
| STCtau_r=3, Heterogeneous ML optimizers=true2026.06 | 6.98 | 11.1 | 15.14 | — |