Table 3.

The architecture of the presented CNN method.

Layer	Type	Properties
1	Feature input	167 × 1 × 1 images
2	Convolution	16 (5 × 5) convolutions with stride [1]
3	ReLU	F(x) = max(0, x)
4	Fully connected	384 fully connected layer
5	Fully connected	384 fully connected layer
6	Fully connected	Four fully connected layer
7	SoftMax	$\sigma {\left(x\right)}_i=\frac{e^{x_i}}{{\sum }_{j=1}^K{e}^{x_j}},i=1,\cdots ,Kx=\left(x_1,\cdots ,x_K\right)$
8	Classification output	For multiclass grouping problems with mutually exclusive groups, the crossentropy loss