NeurIPS 2020

Neural Architecture Generator Optimization

Meta Review

This paper initially got a borderline recommendation (6,5,7). The reviewers agree that this paper gives interesting findings and the idea is new -- it targets at how to generate search space. However, reviewers have some questions on the experiment results. The authors give good response and address these questions well. The ratings are increased to 7,7,6. All the reviewers give positive recommendation. AC agrees.