IPCADP-Equalizer: An Improved Multibalance Privacy Preservation Scheme against Backdoor Attacks in Federated Learning

Algorithm 1

Normal user-level differential algorithm.

(1)	Input: The model parameter , initial limit threshold C, Client sample proportion q, Total communication rounds T, Parameters of each client corresponding to LDP (), Attenuation coefficient γ
(2)	Output: Global model
(3)	for each round t = 0, …, T-1 do
(4)	Select clients CL from
(5)	for each client k in CL do
(6)
(7)
(8)	()
(9)
(10)
(11)
(12)	end
(13)
(14)	end