NeurIPS 2020

Learning of Discrete Graphical Models with Neural Networks

Meta Review

The discussion has converged to a positive evaluation by the reviewers, in particular because the authors' response clarified some important points. In view of that, the authors are strongly invited to take the feedback on board for the final version. There have been some clear points for improvement (position wrt SOTA - comparing only to GRISE has been a point, guarantees or lack of them, attempts to discuss the generality and broader impact beyond learning a structure, etc).