Table 9.

Tested hyperparameters in network layers and optimal values generated via the Grid Search Method [51].

Layers	Number of Tested Units or Filters	Optimal Value	Filter Size	Parameters	Tested Values	Optimal Value
Embedding layer	(50, 100, 150)	50	-	Epochs	50	50
Initial CNNs	(32, 64, 128, 256)	32	{1, 2, …, 9}	Learning rate	(0.005, 0.008, 0.01, 0.02)	0.008
Dilated CNN	(32, 64, 128, 256)	32	3	Batch size	(32, 64, 128, 256)	64
Bottlenecked CNN	(32, 64, 128, 256)	32	1	Decay rate	$({10}^{-7}, {10}^{-8}, {10}^{-9}$)	${10}^{-7}$
Final CNNs	(32, 64, 128, 256)	32	{11, 13, 15}	Early stopping value	(3, 4, 5, 6)	5
FC	(64, 128, 256, 512)	128	-	-	-	-
MaxPooling	(2, 3, 5, 7)	3	-	-	-	-