Image Classification on CIFAR100 (train)
0Communication (Gbit)Local
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Local2024.06 | 0 | — | 1.5384 | |
| SpaFL2024.06 | 4.608 | — | 1.2894 | |
| FedP32024.06 | 2,682.6 | — | 4.9384 | |
| Fjord2024.06 | 5,251.4 | — | 4.1274 | |
| FedSpa2024.06 | 5,342.2 | — | 9.275 | |
| HeteroFL2024.06 | 5,356.4 | — | 4.3634 | |
| FedPM2024.06 | 5,506.1 | — | 5.423 | |
| FedAvg2024.06 | 10,712 | — | 8.7289 | |
| AdaBeliefOptimizer=AdaBelief, Batch size=1024, Backbone=ResNet-182026.02 | — | 85.89 | — | |
| AdaBeliefBackbone=ResNet-18, Batch size=1282026.02 | — | 81.22 | — | |
| AdamOptimizer=Adam, Batch size=1024, Backbone=ResNet-182026.02 | — | 83.15 | — | |
| AdamBackbone=ResNet-18, Batch size=1282026.02 | — | 75.65 | — | |
| LaPropOptimizer=LaProp, Batch size=1024, Backbone=ResNet-182026.02 | — | 83.05 | — | |
| LaPropBackbone=ResNet-18, Batch size=1282026.02 | — | 75.59 | — | |
| MVN-GradOptimizer=MVN-Grad, Batch size=1024, Backbone=ResNet-182026.02 | — | 85.89 | — | |
| MVN-GradBackbone=ResNet-18, Batch size=1282026.02 | — | 81.26 | — |