Table 3.

Properties of the Deep Neural Network models.

DNN Models	Activation Function	Activation Function in the Output Layer	Loss Function	Optimization Algorithm	Epochs and Batch Size
DNN_Model_1 3 hidden layers, with 150 neurons on each hidden	Rectified Linear Unit (RELU)			Adam gradient descent
layer DNN_Model_2 3 hidden layers, with 300 neurons on each hidden		2-choice scale Sigmoid activation function	2-choice scale Binary crossentropy		1000 epochs for training
layer DNN_Model_3 6 hidden layers, with 150 neurons on each hidden		4-choice scale Softmax activation function	4-choice scale Categorical crossentropy and one-hot encoding		Batch size of 20
layer 6 hidden layers, with 300 neurons on each hidden layer