Disentangled Representation Learning