Table 2.

Overview of all layers and model parameters of the proposed CNN architecture. The output shape is given for one batch.

Layer Type	Hyperparameter	Output Shape	# of Parameters
1D Convolution	filter${}_1$: 256, kernelsize: 3	(398, 256)	4864
Max-pooling	poolsize: 2	(199, 256)	0
1D Convolution	filter${}_2$: 128, kernelsize: 3	(197, 128)	98,432
Dropout	dropout${}_1$: 0.30228	(197, 128)	0
Max-pooling	poolsize: 2	(98, 128)	0
1D Convolution	filter${}_3$: 16, kernelsize: 3	(96, 16)	6160
Dropout	dropout${}_2$: 0.03576	(96, 16)	0
Max-pooling	poolsize: 2	(48, 16)	0
Fully-connected		(768)	0
Dropout	dropout${}_3$: 0.43372	(768)	0
Dense		(3)	2307