Disentangling higher level generative factors as disjoint latent dimensions offer several benefits such as ease of deriving invariant representations, targeted data augmentation with style transfer, better interpretability ...