Multiscale Invertible Generative Networks
               for High-Dimensional Bayesian Inference

                 Shumao Zhang 1 Pengchuan Zhang 2 Thomas Y. Hou 1

             Abstract               where L is the likelihood given as
                                       L(y|x) = N (y  F(x); 0, Γ) ,        (2)
    We propose a Multiscale Invertible Generative      which is the density of ε = y  F(x), and Z is some
    Network (MsIGN) and associated training algo-      normalizing constant t ...