This post is a short summary of what needs to be known, and slightly more, to fully get the grasp of a series of not so recent papers (2017-2018) which have tried to exploit the mathematical theory of Optimal Transport to compute multilingual word embeddings.

Continue reading