Genetic CFL: Hyperparameter Optimization in Clustered Federated Learning

Algorithm 1

Initial broadcast round and clustering.

	n = Number of Clients
	: learning rate
	: learning rate list
(1)	procedure Server
(2)	: Server Model Initialization
(3)	Initialize with learning rates
(4)	Broadcast (, random ())
(5)	procedure Client
(6)
(7)	whiledo
(8)
(9)	whiledo
(10)	Train on
(11)
(12)	j = j + 1
(13)	i = i + 1
(14)	return
(15)	procedure Server
(16)
(17)	Initialize DBSCAN Clustering Algorithm
(18)
(19)
(20)
(21)	clusters = model.fit_predict ()