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

Table 7

Recurrent neural network parameters.

Parameter	Description	Default
units	The number of RNN units or an output dimension	Positive integer
Activation	Activation function name	relu, adam, tanh
drop_out	The input linear transformation of the neuron opening ratio	Ratio between 0 and 1
recurrent_dropout	The neuron disconnection ratio of the cyclic state linear transformation	Ratio between 0 and 1
return_sequences	The return of the last output of the output sequence	True, False
go_backwards	Reverses the input sequence and returns the reversed sequence	True, False
Stateful	The final state of a sample being used as the initial state for the next batch	True, False
input_dim	When using the layer as the first layer of the model, specify the value	Positive integer
input_length	The length of the input sequence, or specify this parameter	Positive integer
merge_mode	Combination of forward and backward RNN outputs	sum, mul, none