Table 2. Architectures and hyperparameters tested on the evaluation of deep learning classification models.

Parameters	Layers information	Dropout rate
CNN	1D-Conv (32, filter = 3, kernel = 0) $\to$ MaxPooling (2) $\to$ Flatten $\to$ Dense (64) $\to$ SoftMax	–
LSTM	LSTM (100) $\to$ Dense (100) $\to$ SoftMax	0.5
GRU	GRU (64) $\to$ GRU (32) $\to$ Dense (64) $\to$ SoftMax	–
CNN-LSTM arch. 1	1D-Conv (filter = 16, kernel = 5) $\to$ 2 * 1D-Conv (filter = 64, kernel = 3) $\to$ MaxPooling (2) $\to$ Flatten, LSTM (20) $\to$ Flatten $\to$ Dense (20) $\to$ SoftMax	0.5
CNN-LSTM arch. 2	1D-Conv (filter = 64, kernel = 3) $\to$ Flatten $\to$ LSTM (50) $\to$ Flatten $\to$ SoftMax	0.5