Table 4.

Saturated performances of stochastic channel-based federated learning with pruning when the total pruned proportion was fixed and the pruning rate for each training loop changed.

Pruning rate/loop	AUC-ROCa	AUC-PRb
10%	0.9765	0.9661
20%	0.9730	0.9568
30%	0.9763	0.9662
40%	0.9693	0.9465
50%	0.9769	0.9663

^aAUC-ROC: area under the receiver operating characteristic curve.

^bAUC-PR: area under the precision-recall curve.