Table 1. Architectures of the SC-CNN model.

Layer name	Output size	Kernel size	Pad	Stride	Dropout
conv1	32x56x56	7x7	3	2
max pool	32x28x28	3x3	2
res-conv1	128x28x28	1x1	0	1
		3x3	1	1
		1x1	0	1
res-conv2	256x14x14	1x1	0	2
		3x3	1	1
		1x1	0	1
res-conv3	512x7x7	1x1	0	2
		3x3	1	1
		1x1	0	1
average pool	512x1x1	7x7	0
fc1	500				0.5
fc2	100				0.5
fc3	25				0.5