Variational Auto-Encoders (VAE ... aims to condense the image information into a reduced set of low-level descriptors, called latent variables. The second part, called decoder, aims to use the latent ...