Utility Optimization of Federated Learning with Differential Privacy

Algorithm 1

A-DPFL.

	Input: Dataset , initial , maxRound, , number of client chosen , , learning rate , batch size , clipping threshold , , , noise scale adjusting , adjusting threshold ,
	Output: Global model weight
(1)
(2)
(3)
(4)	whiledo
(5)	= Client choose()//2
(6)	Broadcast()
(7)	fordo
(8)	//Algorithm 3
(9)	Upload()
(10)	end
(11)	//
(12)	ifthen
(13)	//Algorithm 4
(14)	end
(15)
(16)	end
(17)	return