Image Classification on EMNIST (Non-IID, alpha=0.8)
86.91AccuracyFLEX-MoE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FLEX-MoEIndicator=Loss-driven, Client count C=20, Expert count E=8, Expert selection range kc=[2, 6]2025.12 | 86.91 | 0.0041 | 18,482 | |
| GreedyIndicator=Acc-driven, Client count C=20, Expert count E=8, Expert selection range kc=[2, 6]2025.12 | 86.18 | 0.2191 | 1,036,969 | |
| GreedyIndicator=Loss-driven, Client count C=20, Expert count E=8, Expert selection range kc=[2, 6]2025.12 | 85.06 | 0.2093 | 992,532 | |
| FLEX-MoEIndicator=Acc-driven, Client count C=20, Expert count E=8, Expert selection range kc=[2, 6]2025.12 | 83.88 | 0.0038 | 14,688 | |
| RandomIndicator=None, Client count C=20, Expert count E=8, Expert selection range kc=[2, 6]2025.12 | 83.26 | 0.0305 | 190,274 |