Characterizing and Avoiding Problematic Global Optima of Variational Autoencoders

Open in new window