Image Classification on Tiny-ImageNet (Base Top-1 Acc, FedSPC Top-1 Acc, Delta)
65.94Base Top-1 AccuracyFedBABU
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| FedBABUalpha (α)=0.01, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 65.94 | 76.52 | 10.58 | |
| FedPeralpha (α)=0.01, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 65.08 | 75.54 | 10.46 | |
| FedBABUalpha (α)=0.1, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 62.6 | 64.41 | 1.81 | |
| Dittoalpha (α)=0.01, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 61.61 | 68.22 | 6.61 | |
| LG-FedAvgalpha (α)=0.01, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 60.61 | 68.18 | 7.57 | |
| FedPeralpha (α)=0.1, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 60.03 | 59.69 | -0.34 | |
| Dittoalpha (α)=0.1, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 59.39 | 62.88 | 3.49 | |
| FedRepalpha (α)=0.01, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 56.53 | 74.82 | 18.29 | |
| FedRepalpha (α)=0.1, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 55.46 | 61.1 | 5.64 | |
| FedBABUalpha (α)=1.0, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 55.45 | 59.57 | 4.12 | |
| LG-FedAvgalpha (α)=0.1, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 50.6 | 51.55 | 0.95 | |
| Dittoalpha (α)=1.0, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 50.24 | 58.05 | 7.82 | |
| FedPeralpha (α)=1.0, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 47.2 | 48.5 | 1.3 | |
| FedRepalpha (α)=1.0, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 44.6 | 50.65 | 6.05 | |
| LG-FedAvgalpha (α)=1.0, Backbone=ResNet-34, Communication rounds (r)=1002026.06 | 30.4 | 32.32 | 1.92 |