Impact of COVID-19 on Forecasting Stock Prices: An Integration of Stationary Wavelet Transform and Bidirectional Long Short-Term Memory

Table 3

Hyperparameter values used in the model training process. The first column represents hyperparameter name, and in the second column, possible parameters are shown.

Hyperparameter	Possible parameters
BDLSTM—hidden layer size	(32), (64), (32, 32), (64, 32), (64, 64), (32, 32, 32), (64, 32, 32), (64, 32, 64), (64, 64, 64)
FC—hidden layer size	(12), (24), (12, 12), (24, 12), (24, 24), (12, 12, 12), (24, 12, 12), (24, 12, 24), (24, 24, 24)
Activation function	ReLU, ELU, tanh, identity
Optimizer	Adam, RMSprop
Learning rate	0.0001, 0.001, 0.01
Learning rate decay	1e − 7, 1e − 6, 1e − 5
Regularization parameter L2	0.0001, 0.001, 0.01