NeurIPS 2020

Locally Differentially Private (Contextual) Bandits Learning

Meta Review

After discussion, all reviewers agreed that there are interesting and novel contributions in this paper, and that it meets the bar for acceptance. The main outstanding concern is clarity, in particular with regard to how the novel LDP setting relates to the central DP setting for contextual bandits. (This is discussed in Review 3 and acknowledged in the author feedback.) More generally, I hope the authors can make the writing a bit easier to follow for the final version, following the suggestions of the reviewers.