Table 8. Best value of hyperparameters tuned in deep neural networks.

Parameter	Best value
	FNN	CNN
Batch size	20	10
Epochs	100	10
Training optimization algorithm	Adagrad	SGD
Learn rate	0.2	0.1
Momentum	0	0
Network weight initialization	Normal	Glorot_normal
Neuron activation function	Softsign	Linear
Weight constraint	3	1
Dropout regularization	0	0.4