Review for NeurIPS paper: Estimating decision tree learnability with polylogarithmic sample complexity

Additional Feedback: The paper is not interesting enough for a competitive conference. It is good to have these results in the literature, but I suggest to send it to a journal. Having read the reviews, and following the discussion, I still think that this does not below in a competitive conference. Indeed, as the authors stress in their response, the power of the result is due to the specific algorithm developed here. Nevertheless, I cannot be excited by it, given the monotonicity assumption and the fact that it applies only to the uniform distribution setting. I agree that it's an interesting result, but I think that it's not interesting enough nor important enough for a top conference.