Neural approximation of Wasserstein distance via a universal architecture for symmetric and factorwise group invariant functions

Open in new window