[Retracted] An ECG Heartbeat Classification Method Based on Deep Convolutional Neural Network

Table 3

The structure of the proposed CNN classifier.

	Type	Biases	Kernel size	Stride	Output size
Layer 1	Convolution	16	3	1	(185, 16)
Layer 2	Convolution	16	3	1	(183, 16)
Layer 3	Max pooling	—	2	2	(91, 16)
Layer 4	Convolution	32	3	1	(89, 32)
Layer 5	Convolution	32	3	1	(87, 32)
Layer 6	Max pooling	—	2	2	(43, 32)
Layer 7	Convolution	64	3	1	(41, 64)
Layer 8	Convolution	64	3	1	(39, 64)
Layer 9	Max pooling	—	2	2	(19, 64)
Layer 10	Convolution	128	3	1	(17, 128)
Layer 11	Convolution	128	3	1	(15, 128)
Layer 12	Convolution	128	3	1	(13, 128)
Layer 13	Max pooling	—	2	2	(6, 128)
Layer 14	Flatten	—	—	—	(768)
Layer 15	Fully connected	30	—	—	(30)
Layer 16	Fully connected	5	—	—	(5)