Benchmarks, Algorithms, and Metrics for Hierarchical Disentanglement

                     Andrew Slavin Ross 1 Finale Doshi-Velez 1

              Abstract              able (Locatello et al., 2018), and it seems unlikely that con-
    In representation learning, there has been re-      tinuous, factorized, fixed-dimensional representations are
    cent interest in developing algorithms to disen-     the optimal choice for modeling many real-world generative
    tangle the ground-truth generati ...