NeurIPS 2020

Online Learning in Contextual Bandits using Gated Linear Networks


Meta Review

The paper has received mixed reviews, and the mixed opinion of reviewers was reflected in the discussion. The main strengths and weaknesses of the paper are consistent across reviewers however. The reviewers appreciated that the paper provides progress in important practical setting and the results seem to be correct. The main concerns are that the paper could be difficult to read for those unfamiliar with GLN, that the experimental results are not fully justified/explained. I find the experimental setup inline with the standard in this field, and hope that the authors can provide some more explanations based on reviewer suggestions to make the camera-ready version more readable.