Skip to main content
. 2023 Oct 26;14:1176753. doi: 10.3389/fphys.2023.1176753

TABLE 3.

CNN AutoEncoder architecture.

Layers Output shape
Encoder
Input layer (None, 100)
Convolution layer (None,100,32)
MaxPooling (None,50,32)
Convolution layer (None,50,32)
MaxPooling (None,25,32)
Convolution layer (None,25,32)
MaxPooling (None,13,32)
Flatten (None,416)
Dense (None,8)
Decoder
Dense (None,8)
Dense (None,416)
Reshape (None,13,32)
Convolution layer (None,26,32)
Cropping (None,25,32)
UpSampling (None,50,32)
Convolution layer (None,50,32)
UpSampling (None,100,32)
Convolution layer (None,100,32)
Flatten (None,3,200)
Dense (None,100)