VAE; representational similarity; representation learning; deep learning; variational autoencoders