Image Classification on CIFAR-100 10%-100 devices
59.96Acc (Dirichlet α=0.5)SCAFFOLD
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| SCAFFOLDDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 59.96 | 274 | 50.59 | 450 | 55.78 | 368 | 47.05 | 534 | |
| FedVSSAMDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 58.78 | 237 | 51.22 | 328 | 56.87 | 258 | 50.27 | 337 | |
| FedSMOODevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 56.87 | 279 | 49.12 | 396 | 55.04 | 330 | 48.6 | 463 | |
| FedAdamDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 54.8 | 155 | 48.29 | 276 | 50.38 | 376 | 43.36 | 637 | |
| FedSAMDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 52.22 | 439 | 43.5 | 688 | 47.52 | 530 | 39.84 | — | |
| FedACGDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 51.66 | 106 | 46.3 | 161 | 51.85 | 124 | 46.18 | 232 | |
| FedGAMMADevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 51.39 | 511 | 38.91 | — | 45.29 | 638 | 35.57 | — | |
| FedDynDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 50.84 | 177 | 44.83 | 307 | 50.4 | 209 | 44.72 | 377 | |
| FedCMDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 48.51 | 145 | 44.63 | 254 | 49.71 | 226 | 45.32 | 343 | |
| MoFedSAMDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 47.14 | 576 | 49.69 | 450 | 54.49 | 360 | 45.6 | 589 | |
| FedProxDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 46.8 | 313 | 37 | — | 41.6 | 517 | 36.75 | — | |
| FedLESAMDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 46.58 | 308 | 35.85 | — | 43.19 | 458 | 36.33 | — | |
| FedAvgDevice Sampling=10%-100 devices, Training Rounds=8002026.05 | 46.52 | 313 | 36.76 | — | 42.59 | 486 | 36.67 | — |