Gradient Leakage Attack on Tiny ImageNet
1,184.02Robbing the Fed OverheadRobbing the Fed
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Robbing the FedBackbone=VGG-11, Model size (MB)=506.8334, Number of clients=1000, Batch size=642023.03 | 1,184.02 | — | |
| Robbing the FedBackbone=Inception v3, Model size (MB)=103.612, Number of clients=1000, Batch size=642023.03 | 5,791.79 | — | |
| Robbing the FedBackbone=ResNet-50, Model size (MB)=97.4923, Number of clients=1000, Batch size=642023.03 | 6,155.35 | — | |
| Robbing the FedBackbone=ResNet-18, Model size (MB)=44.5919, Number of clients=1000, Batch size=642023.03 | 13,457.57 | — | |
| Robbing the FedBackbone=MobileNet v3 (L), Model size (MB)=20.9161, Number of clients=1000, Batch size=642023.03 | 28,690.76 | — | |
| Sparse MANDRAKEBackbone=MobileNet v3 (L), Model size (MB)=20.9161, Number of clients=1000, Batch size=642023.03 | — | 87.65 | |
| Sparse MANDRAKEBackbone=ResNet-18, Model size (MB)=44.5919, Number of clients=1000, Batch size=642023.03 | — | 41.11 | |
| Sparse MANDRAKEBackbone=ResNet-50, Model size (MB)=97.4923, Number of clients=1000, Batch size=642023.03 | — | 18.8 | |
| Sparse MANDRAKEBackbone=Inception v3, Model size (MB)=103.612, Number of clients=1000, Batch size=642023.03 | — | 17.69 | |
| Sparse MANDRAKEBackbone=VGG-11, Model size (MB)=506.8334, Number of clients=1000, Batch size=642023.03 | — | 3.62 |