So far, we have explored vanilla and multilayer autoencoders. In this section, we will see how convolutional autoencoders work in reconstructing the original images from a lower-dimensional vector.
Convolutional autoencoders look as follows:
Essentially, a convolutional autoencoder reconstructs the input with more hidden layers in its network where the hidden layers consist of convolution, pooling, and upsampling the downsampled image.
Similar to a multilayer autoencoder, a convolutional autoencoder differs from other types of autoencoder in its model architecture. In the following code, we will define the model ...
