Table 2.

Investigated architectural hyper-parameters for each architectural style.

Parameters		Architecture style
		Standard	Residual	Dense
Layers		1, 2, 3, 4	1, 2, 3, 4	1, 2, 3
Convolution blocks	Initial	1, 2, 3	1, 2, 3, 4	2, 3
	Increase rate	0, 1, 2	0, 1, 2	0, 1
	Maximum	4	4	4
Filters	Initial	8, 16, 32	32	8, 12, 16, 32
	Growth Rate	–	–	0, 4
	Maximum	32, 64, 128	128	128, inf

Note that the number of filters doubled after each pooling. “Growth Rate” is unique to the Dense network.