Dermoscopic Image Classification Using Deep Belief Learning Network Architecture

Table 2

Proposed DBL network architecture.

Layer	Layer type	No. vol	Kernel size	Stride
1	Convolution 1+ReSF	96		4
2	Probabilistic pooling			1
3	Convolution 2+ReSF	512		1
4	Probabilistic pooling			1
5	Convolution 3+ReSF	384		1
6	Convolution 4+ReSF	384		1
7	Convolution 5+ReSF	256		1
8	Probabilistic pooling			1
9	3× fully connected	4096 hidden units