Automatic Sleep Stage Classification Based on Convolutional Neural Network and Fine-Grained Segments

Table 2

Details of CNN architectural layers.

Layer	Layer type	Units	Units type	Kernel size	Stride	Output dimension
: Input						(N,T)
Reshape						(N,T,1)
	Convolutional	20	Relu	(1,5)	(1,1)	(N,T,20)
	Max-pooling			(1,10)	(1,10)	(N,T,20)
	Convolutional	30	Relu	(1,5)	(1,1)	(N,T,600)
	Max-pooling		Relu	(1,10)	(1,10)	(N,T,600)
	Full-connected	500	Relu			500
	softmax	5				5