Table 1.

Convolutional neural network (CNN) architecture variations for tested models.

Layers	Layer Parameters	Model 1	Model 2	Model 3	Model 4	Model 5
	Filters	16	16	8	8	16
Convolutional Layer	Kernel Size	$4\times 4$	$4\times 4$	$4\times 4$	$4\times 4$	$4\times 4$
	Activation Function	ReLu	ReLu	ReLu	ReLu	ELU
Max Pooling	Kernel Size	$2\times 2$	$2\times 2$	$2\times 2$	$2\times 2$	$2\times 2$
Dropout		0.2	0.2	0.2	0.2	0.2
	Filters	16	16	8	8	16
Convolutional Layer	Kernel Size	$5\times 5$	$5\times 5$	$5\times 5$	$5\times 5$	$5\times 5$
	Activation Function	ReLu	ReLu	ReLu	ReLu	ELU
Max Pooling	Kernel Size	$2\times 2$	$2\times 2$	$2\times 2$	$2\times 2$	$2\times 2$
Dropout		0.1	0.1	0.1	0.1	0.1
	Filters	16	16	8	8	16
Convolutional Layer	Kernel Size	$6\times 6$	$6\times 6$	$6\times 6$	$6\times 6$	$6\times 6$
	Activation Function	ReLu	ReLu	ReLu	ReLu	ELU
Max Pooling	Kernel Size	$2\times 2$	$2\times 2$	$2\times 2$	$2\times 2$	$2\times 2$
Dense	Neurons	64	64	64	64	64
	Activation Function	ReLu	ReLu	ReLu	ReLu	ELU
Dense	Neurons	128	32	32	128	128
	Activation Function	ReLu	ReLu	ReLu	ReLu	ELU
Dense	Neurons	64	16	16	64	64
	Activation Function	ReLu	ReLu	ReLu	ReLu	ELU
Dense	Neurons	4	4	4	4	4
	Activation Function	ReLu	ReLu	ReLu	ReLu	ELU
Test Loss		0.2413	0.2459	0.1891	0.2040	0.2145
Test Accuracy		0.9440	0.9397	0.9483	0.9586	0.9570