NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:71
Title:Adversarial Examples Are Not Bugs, They Are Features


		
A very good paper, on which all reviewers early agreed on its merits and originality. I strongly suggest the the authors follow on reviews and implement the suggestions they proposed to do, in particular those that allowed score increase (R#2) and, as much as possible, those suggested in L37-L40 of their rebuttal, at least as additional remark in the main file if the space allows it (eventually amending the conclusion to reflect some of the work mentioned). It could represent a great additional appeal for the interested reader and would surely contribute to prepare some feedback and questions for the conference.