Delta-DAGMM: A Free Rider Attack Detection Model in Horizontal Federated Learning

Algorithm 1

Delta-DAGMM algorithm.

	Input: n // number of training rounds, m // number of participants, // local updated model parameter
	Output: atkList // list of Free-rider attack detection results
(1)	Initialize the global model and the list of attack detection results atkList
(2)	for t = 1 to n do // t denotes the current training round
(3)	Parameter server sends the global model to all participants
(4)	// Get the energies of samples calculated by Delta-DAGMM
(5)	for i = 1 to m do
(6)	// Gets the participant local update model increment
(7)
(8)	// Process the model increment
(9)
(10)	// Input the processed model increment as a sample
(11)
(12)	end for
(13)	// Free-rider attack detection
(14)
(15)	for i = 1 to m do
(16)	// Set threshold
(17)	if then
(18)
(19)	end for
(20)	// Update the global model
(21)
(22)	end for
(23)	return atkList
(24)	end for