Supervised Tree-Wasserstein Distance

                  Yuki Takezawa 1 2 Ryoma Sato 1 2 Makoto Yamada 1 2

              Abstract               the cost of the optimal transport between two documents in
    To measure the similarity of documents, the        the word embedding space. WMD has been used for doc-
    Wasserstein distance is a powerful tool, but it      ument classification tasks and has achieved high k-nearest
    requires a high computational cost. Recently,       neighbors (kNN)  ...