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

Feb-8-2026, 16:35:39 GMT–Neural Information Processing Systems 

However, functions on such complex objects (e.g., point sets and graphs) are often