A DDoS-Attack Detection Method Oriented to the Blockchain Network Layer

Table 1

Parameters selected for the CMCNN model.

Parameter	Description	Value
M	Batch training size	50
K	Dimension of input traffic features	163
Channel	Number of model channels	2
ρ′ m	Average activation degree of the SSAE layer	0.52
λ	L2 weight regularisation loss	0.49
β	Coefficient of sparse control weight	0.89
α	Momentum parameter	0.97
ξ	Learning rate adjustment parameter	50
ε	Threshold of changes in model accuracy rate	0.05
μ	Leakage parameter of Leaky ReLU	0.1
ηmin	Min value of the learning rate	0.5
ηmax	Max value of the learning rate	0.99
Niter	Total number of model iterations	1200