NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:1357
Title:Learning to Control Self-Assembling Morphologies: A Study of Generalization via Modularity

The reviewers praise the paper's goal of integrating a model for dynamic morphologies and deep RL, as well as the clarity of its writing. Several reviewers bring up concerns around the issue of baselines. The reviewers ask for more comparisons with baselines in situations where both the proposed method and the baselines work, more empirical evidence about the reproducability of the proposed method's performance, and evidence for the strength of the baselines. Some of the reviewers' concerns were not addressed by the authors in their response. I strongly encourage the authors address these concerns in the camera ready version of the manuscript, either in the writing of the paper or with new experiments, or ideally with both. I look forward to reading the camera ready version of the manuscript with these considerations in place.