Image Classification on CIFAR-10 (test) (Accuracy and MIA Accuracy)
30.16AccuracyERIS
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ERISBackbone=ResNet92026.02 | 30.16 | 62.72 | |
| FedAvgBackbone=ResNet92026.02 | 29.83 | 63.52 | |
| Min. LeakageBackbone=ResNet92026.02 | 29.8 | 61.92 | |
| PriPruneBackbone=ResNet9, Pruning rate (p)=0.012026.02 | 21.3 | 59.73 | |
| FedAvg (ε, δ)-LDPBackbone=ResNet9, Privacy budget (epsilon)=102026.02 | 15.13 | 58.83 | |
| SoteriaFL (ε, δ)Backbone=ResNet9, Privacy budget (epsilon)=102026.02 | 14.46 | 59.07 | |
| ShatterBackbone=ResNet92026.02 | 11.63 | 62.19 | |
| PriPruneBackbone=ResNet9, Pruning rate (p)=0.052026.02 | 11.51 | 57.71 | |
| PriPruneBackbone=ResNet9, Pruning rate (p)=0.12026.02 | 10.98 | 55.53 |