After the discussion phase, there is a consensus among the knowledgeable reviewers that this is a good paper that warrants acceptance. The paper is therefore accepted as poster. However, there are a few points that I urge the authors to take into consideration when preparing the camera-ready version. In my opinion, the reviews, especially that of R3, are quite thorough already, so I hope the authors will consider some of the important suggested changes. Hence, I only emphasize the important points here. First, please improve the readability and presentation of the paper and make sure that it is accessible to the machine learning folks. Second, please re-consider the title of the paper. The title should clearly reflect the content of the paper. Please, avoid the title that is too cryptic (but it's also fine if the authors decide to keep the title as is).