Learning representations that disentangle the underlying factors of variability in data is an intuitive precursor to AI with human-like reasoning. Consequently, it has been the object of many efforts of the machine learning community. This work takes a step further in this direction by addressing the scenario where →