Handwritten character recognition on EMNIST 8 (test)
83.5Average AccuracyFedEM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FedEMModel=2-layer CNN + 2-layer FFN, M=32021.08 | 83.5 | 76.6 | |
| pFedMeModel=2-layer CNN + 2-layer FFN2021.08 | 83.3 | 76.4 | |
| FedAvg+Model=2-layer CNN + 2-layer FFN2021.08 | 83.1 | 75.8 | |
| FedProxModel=2-layer CNN + 2-layer FFN2021.08 | 83 | 75.4 | |
| Clustered FLModel=2-layer CNN + 2-layer FFN2021.08 | 82.7 | 75 | |
| FedAvgModel=2-layer CNN + 2-layer FFN2021.08 | 82.6 | 75 | |
| LocalModel=2-layer CNN + 2-layer FFN2021.08 | 71.9 | 64.3 |