Table 2 The proposed NN architecture for elementwise classification of input MII data.

From: Segmentation of gastroesophageal reflux events using a semi-U-Net architecture with 1D/2D CNNs

Layer

Layer type

#Channels

Unit type

Kernel size

Stride

Padding

Output

Encoder

       

Input*

 

1

    

(1, 6000, 6)

C1**

Conv 2-d

8

Relu

(211, 6)

 

(105, 0)

(8, 6000)

P

Pooling

 

Max

 

(2, 1)

 

(8, 3000)

C2

Conv 1-d

16

Relu

21

 

10

(16, 3000)

P

Pooling

 

Max

 

2

 

(16,1500)

Decoder

       

D1

Transpose

Conv 1-d

8

Relu

2

2

 

(8, 3000)

D2

Transpose

Conv 1-d

1

 

2

2

 

(1, 6000)

Output

Softmax

 

Sigmoid

   

(1, 6000)

  1. *, **: Attribute functions unsqueeze(1) and squeeze(3) were used before input and after C1 layers, respectively.
Back to article page