NeurIPS 2020

Provably Efficient Neural GTD for Off-Policy Learning

Meta Review

This paper generated substantial discussion from the reviewers. Reviewer 1's points of lack of contextualization are well-taken by the other reviewers. That said, the meta-reviewer (in consultation with the Senior Area Chair) agrees that the theoretical contribution will be of interest to the NeurIPS community, and the clarity & sharpness of the authors' response suggests the authors are quite capable of revising the paper to more clearly discuss context and articulate their contribution. As such, the metareviewer is recommending accept.