Preliminary Study on the Efficient Electrohysterogram Segments for Recognizing Uterine Contractions with Convolutional Neural Networks

Table 1

Detailed parameters used for all the layers of the CNN model.

Layer	Type	Kernel size	Other layer parameters
1	Conv + ReLU	27	Strides = 5, num_output = 96
2	Max-pooling	2	Strides = 3
3	Conv + LRN + ReLU	2	Strides = 1, local_size = 5, α = 0.0001, β = 0.75, num_output = 256
4	Max-pooling	2	Strides = 2
5	Conv + LRN + ReLU	3	Strides = 1, num_output = 384, local_size = 5, α = 0.0001, β = 0.75
6	Conv + ReLU	3	Strides = 1, pad = 2, num_output = 384
7	Conv + ReLU	3	Strides = 1, pad = 2, num_output = 256
8	Max-pooling	3	Strides = 2
9	FC + ReLU + dropout		Dropout_ratio = 0.5, num_output = 4096
10	FC + ReLU + dropout		Dropout_ratio = 0.5, num_output = 4096
11	FC		num_output = 2, activation = softmax

Pad is the padding number with leading zeros.