Exploration of Stock Portfolio Investment Construction Using Deep Learning Neural Network

Table 1

Parameter setting of DL NN model.

Type	Value
Network node	75% of total points
Initial weight	−0.5∼0.5
Minimum training rate	0.9
Dynamic parameter	0.6∼0.8
Allowable error	0.001∼0.00001
Number of iterations	1000
Sigmoid parameters	0.9∼1