Deep Learning-Based Surface Nerve Electromyography Data of E-Health Electroacupuncture in Treatment of Peripheral Facial Paralysis

Table 2

Hyperparameter settings of the preaggregation network.

Hidden layer		Hyperparameter		Hidden layer		Hyperparameter
Name	Type	Name	Setting	Name	Type	Name	Setting
First layer	Convolutional layer	Size of convolution kernel		Third layer	Local connection layer	Number of convolution kernels	64
First layer	Convolutional layer	Number of convolution kernels	64	Fourth layer	Fully connected layer	Number of hidden units	512
Second layer	Local connection layer	Size of convolution kernel		Fifth layer	Fully connected layer	Number of hidden units	512
Second layer	Local connection layer	Number of convolution kernels	64	Sixth layer	G-way fully connected layer	Number of hidden units	Number of facial images to be classified
Third layer	Local connection layer	Size of convolution kernel		Seventh layer	Softmax classification function	—	—