Review for NeurIPS paper: Sample complexity and effective dimension for regression on manifolds

Weaknesses: The authors do not propose a new algorithm in this paper but only establish the theoretical results to reveal the relationship between the intrinsic dimension and the regression on manifolds. The function space concerned is the classic RKHS space using the heat kernel. It's unclear to me whether the author's theoretical results have significantly improved the previous regression or classification algorithms. The author did not design further experiments to illustrate this point. The innovation and contribution of the article are limited and unattractive.