# Invariant Federated Learning

## Dataset: ICU

### Arithmetic Mean

Total training round: 10
Training epochs: 30/Round
Total clients: 38
Feature selection: 256


ROC score: 0.512
Precision: 0.066
Recall: 0.209
F1 Score: 0.101
PR AUC: 0.059

### Geometric Mean

Total training round: 10
Training epochs: 30/Round
Total clients: 38
Feature selection: 256


ROC score: 0.480
Precision: 0.047
Recall: 0.269
F1 Score: 0.081
PR AUC: 0.0466

### And Mask

Total training round: 10
Training epochs: 30/Round
Total clients: 38
Feature selection: 256


ROC score: 0.514
Precision: 0.055
Recall: 0.9013
F1 Score: 0.103
PR AUC: 0.061

## Dataset: MNIST (Binary Classification for 0 and 1)

### Arithmetic Mean

Total training round: 10
Training epochs: 10/Round
Total clients: 10


ROC score: 0.9998141862863036
Precision: 1.0
Recall: 0.9955533596837944
F1 Score: 0.997771725674672
PR AUC: 0.999837476363806

### Geometric Mean

Total training round: 10
Training epochs: 10/Round
Total clients: 10


ROC score: 0.9998477201531499
Precision: 0.5336842105263158
Recall: 1.0
F1 Score: 0.6959505833905285
PR AUC: 0.9998632198426646

### And Mask

Total training round: 10
Training epochs: 10/Round
Total clients: 10


ROC score: 0.9999265585723632
Precision: 0.9975704567541303
Recall: 0.9985408560311284
F1 Score: 0.9980554205153136
PR AUC: 0.9999375948846042