Federated Learning Image Classification on DirtyMNIST
0.434Max r_k(θ)EAGLE
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| EAGLEModel=CNN, λ=5.02026.03 | 0.434 | 0.249 | 81.4 | 0.006 | |
| EAGLEModel=CNN, λ=3.02026.03 | 0.322 | 0.14 | 81.8 | 0.005 | |
| EAGLEModel=CNN, λ=1.02026.03 | -0.002 | -0.506 | 84.3 | 0.034 | |
| AFLModel=CNN2026.03 | -0.016 | -0.822 | 88.2 | 0.083 | |
| q-FFLModel=CNN, q=3.02026.03 | -0.062 | -0.883 | 88.3 | 0.088 | |
| q-FFLModel=CNN, q=5.02026.03 | -0.07 | -0.876 | 87.9 | 0.083 | |
| EAGLEModel=CNN, λ=0.72026.03 | -0.076 | -0.787 | 87.8 | 0.067 | |
| FedAvgModel=CNN2026.03 | -0.083 | -0.877 | 88.9 | 0.081 | |
| q-FFLModel=CNN, q=1.02026.03 | -0.084 | -0.886 | 88.9 | 0.084 | |
| EAGLEModel=CNN, λ=0.52026.03 | -0.094 | -0.869 | 89.3 | 0.082 | |
| EAGLEModel=CNN, λ=0.12026.03 | -0.095 | -0.944 | 89.7 | 0.093 | |
| EAGLEModel=CNN, λ=0.32026.03 | -0.099 | -0.896 | 89.7 | 0.086 |