Projection Robust Wasserstein Barycenters

                     Minhui Huang 1 Shiqian Ma 2 Lifeng Lai 1

             Abstract                However, computing the WB for a set of probability dis-
    Collecting and aggregating information from sev-      tributions is notoriously hard. The hardness comes from
    eral probability measures or histograms is a fun-     two aspects: the representation of the measure support and
    damental task in machine learning. One of the       the potential  ...