Premature Ventricular Contractions’ Detection Based on Active Learning

Table 2

A detailed description of the architecture of the convolutional autoencoder model.

Layer number	Layer name	Kernel size	Output size
0	Input		(−1, 250)
1	Reshape		(−1, 250, 1)
2	Convolution1D	20	(−1, 25, 25)
3	Reshape		(−1, 25, 25, 1)
4	Convolution2D	(1, 20)	(−1, 25, 1, 1)
5	Convolution2D transpose	(1, 20)	(−1, 25, 1, 25)
6	Convolution2D transpose	(20, 1)	(−1, 250, 1, 1)
7	Reshape/output		(−1, 250)