The Design and Implementation of Cardiotocography Signals Classification Algorithm Based on Neural Network

Table 5

Convolution neural network parameters.

Parameter	Description	Default
Filters	The number of filters in the convolution operation	8, 16, 32
kernel_size	The size of the convolution kernel	(3, 3), (5, 5)
Strides	Convolutional step size	(4,4)
Activation	Activation function	relu, adam
pool_size	Pool size	(2,2)
Padding	Whether zero padding	Same, valid
drop_out	The proportion of nonworking neurons	0.25
Optimizier	Optimization function	rmsprop, adagrad
Epochs	The number of complete training samples of the data set	50, 100
batch_size	The number of training samples per iteration	32, 64, 128
data_augmentation	Whether to use data enhancement	True