Image Classification on CIFAR-100 5%-200 devices (Dirichlet & Pathological Robustness)
57.53Accuracy (Dirichlet alpha=0.5)FedVSSAM
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| FedVSSAMDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 57.53 | 248 | 50.45 | 376 | 55.94 | 316 | 50.17 | 398 | |
| FedSMOODevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 54.02 | 289 | 50.13 | 395 | 49.96 | 333 | 46.84 | 472 | |
| SCAFFOLDDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 53.96 | 277 | 47.85 | 484 | 46.27 | 455 | 45.63 | 584 | |
| FedGAMMADevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 52.86 | 466 | 40.09 | 799 | 48.38 | 584 | 36.12 | — | |
| MoFedSAMDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 47.12 | 580 | 48.47 | 513 | 55.16 | 349 | 47.4 | 572 | |
| FedSAMDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 46.99 | 448 | 39.63 | — | 40.75 | 723 | 35.84 | — | |
| FedAdamDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 45.96 | 259 | 43.43 | 494 | 39.65 | — | 41.29 | 720 | |
| FedACGDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 45.77 | 152 | 44.48 | 246 | 45.73 | 240 | 40.31 | 670 | |
| FedCMDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 45.77 | 183 | 47.93 | 269 | 43.11 | 379 | 42.81 | 467 | |
| FedDynDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 35.53 | — | 32.73 | — | 34.6 | — | 30.78 | — | |
| FedAvgDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 32.45 | — | 28.29 | — | 30.39 | — | 27.74 | — | |
| FedLESAMDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 31.97 | — | 34.75 | — | 30.11 | — | 28.24 | — | |
| FedProxDevice Sampling=5%-200 devices, Training Rounds=8002026.05 | 31.56 | — | 33.21 | — | 29.92 | — | 27.18 | — |