CVPR2021-第二封邮件关于rebuttal的要求

Folks,

1) So you have your CVPR reviews...  This note to answer some questions
that have cropped up several times already.

2)  We've had several questions about rebuttal length.  For example,
I got long reviews (five reviews; weird reviews; etc.), can I write a
longer rebuttal?

      The answer is NO, *in all cases*.  Rebuttals should be 1 page pdf max,
       *whatever* your circumstances.

3)  We've had several questions about experiments in rebuttals, and how
we interpret the PAMI-TC policy on experiments in rebuttals.  This
is reproduced below, and can be found at:

 

http://cvpr2021.thecvf.com/node/33#policies

      We interpret this policy as a shield for authors.  If you are asked to
      do more experiments, you may point out the policy as a shield.
      Reviewers should not penalize you for not performing more experiments.   
      Reviewers may (but are not required to) discount additional experiments
      in the rebuttal.  
      Reviewers should not penalize you for having done more experiments, either.

4)  We've had several questions about bad/unreasonable/etc. reviews.  For example,
R1 really didn't understand the paper, etc.

       Don't raise this with PCs;  it's what rebuttals are for.

5) We've had several questions about reviewers requesting paper rewrites, code, etc.

       It's not possible to upload a revised version of the paper at this stage. 
       It's not allowed to provide a link to code in the rebuttal; stick to the 1 page pdf.

best

CVPR 21 PCs

伙计们,

1)你有CVPR评审…这张便条回答了一些问题

这种情况已经出现过好几次了。

2)我们对答辩长度有几个问题。例如, 我得到了很长的评论(5条评论;奇怪的评论;等等)，我可以写一个

再反驳吗?

答案是否定的，*在所有情况下*。反驳应最多1页pdf格式，

* *你的环境。

3)我们有几个关于反驳实验的问题，以及如何反驳

我们在反驳实验中解释了PAMI-TC政策。这

转载如下，并可在以下网址找到:

 

我们将此政策解释为为作者提供的保护。
如果你被要求的话做更多的实验，你可能会指出政策是一个盾牌。

审稿人不应该因为你没有做更多的实验而处罚你。

审稿人可能(但不是必须)对额外的实验打折扣在反驳。

审稿人也不应该因为你做了更多的实验而处罚你。

 

4) 我们有几个关于坏的/不合理的/等等的问题。评论。例如,

R1真的不懂这篇论文，等等。
不要向个人电脑提出这个问题; 这就是反驳的意义所在。
我们有几个关于审稿人要求重写论文、代码等的问题。
目前不可能上传论文的修订版。
不允许在反驳中提供代码链接;坚持使用1页的pdf文件。

Per a passed 2018 PAMI-TC motion, reviewers should not request additional experiments for the rebuttal, or penalize authors for lack of additional experiments. This includes any experiments that involve running code, e.g., to create tables or figures with new results. Examples of such experiments are ablation studies, new baseline results (unless already reported elsewhere), results with additional hyperparameters, or results on new datasets. Authors should not include new experimental results in the rebuttal, and reviewers should discount any such results when making their final recommendation. These also include experiments that were run by authors before the submission deadline but were not included either in the paper or the supplementary material.

Should any evidence of new experiments be found in the rebuttal, the reviewers and Area Chairs have the right to disregard the entire rebuttal. We will encourage ACs/reviewers to ignore rather than penalize additional results if they are reported by authors. ACs are expected to monitor reviewers to make sure authors are not penalized for lack of additional results, even if reviewers asked for them.

根据2018年通过的PAMI-TC动议，审稿人不应要求在反驳时进行额外实验，或因缺乏额外实验而处罚作者。这包括任何涉及运行代码的实验，例如，创建具有新结果的表格或图形。
此类实验的例子有消融研究、新的基线结果(除非已经在其他地方报道)、附加超参数的结果或在新的数据集上的结果。
作者不应该在反驳中包含新的实验结果，而审稿人在做出最终推荐时应该对任何这样的结果打折扣。
这些也包括作者在提交截止日期之前进行的实验，但这些实验既没有包含在论文中，也没有包含在补充材料中。

如果在反驳中发现任何新实验的证据，审稿人和区域主席有权无视整个反驳。
如果作者报告了额外的结果，我们鼓励ACs/审稿人忽略而不是处罚。
ACs被期望监视审稿人，以确保作者不会因为缺少额外的结果而受到惩罚，即使审稿人要求得到这些结果。

Download the CMT app to access submissions and reviews on the move and receive notifications: https://apps.apple.com/us/app/conference-management-toolkit/id1532488001.

To stop receiving conference emails, you can check the 'Do not send me conference email' box from your User Profile.

Microsoft respects your privacy. To learn more, please read our Privacy Statement.

Microsoft Corporation
One Microsoft Way
Redmond, WA 98052