NeurIPS 2020

Generalization bound of globally optimal non-convex neural network training: Transportation map estimation by infinite dimensional Langevin dynamics

Meta Review

Four knowledgeable referees consider this a timely and significant contribution with several methodological insights and novelties, which could be influential. A few concerns are the dense presentation and possible difficulties with interpretation and the role of regularization. I would like to encourage the authors to make an effort to address these points and try to make the contents more accessible in the final manuscript. I am recommending a spotlight accept.