CVPR 2020评审结果放出,旷视南京研究院负责人浅谈学术论文rebuttal
近日,CVPR 2020 论文 review 结果放出,一些读者们是不是开始为 rebuttal 做准备了?在此文章中,旷视南京研究院负责人魏秀参博士介绍了论文 rebuttal 那些事,相信能为读者们提供全新的洞见。
提交 PDF 文档作为 rebuttal(有页数和模版限制,方便提交新表格和插图,如 CVPR/ICCV/ECCV);
提交一段文本作为 rebuttal(有字符数限制,不方便提交新表格和插图,如 AAAI/IJCAI);
提交 PDF 文档作为 rebuttal,同时提交修改后的论文(重提交的文件有类似的政策限制,如 WACV)。
Rebuttal 时,针对不同 review 类型:
喷「novelty」:这类最为麻烦但也是作者们遇到可能最多的意见,作此类 rebuttal 时不妨重新梳理和强调文章的重要贡献,然后澄清并不是 trivial 的简单 combine,再强调一下 motivation 和 intuition,用另一种方式将文章亮点表达出来。同时,可以尝试「围魏救赵」,即:若审稿人针对方法的某个部件提出 novelty 不足,可强调其他部件或整个方法的范式是前所未有的;或 claim 说方法简单有效,思路全然不同;
喷「factual error」:审稿人一旦找出文章的事实性错误,作者不妨大方承认,并表示感谢,同时表示会在 final version 中更正错误;另一种情况是,可能就是因为作者自己没写明白,才使得审稿人错误理解,如此,也可大方承认,说「我们已经修改了这部分描述,实际上是这样做的,并不是你理解的那样,blabla」;
喷「涨点不足」:此类一般有两种,一则无证据的裸喷涨点不足;二则有证据(提供了 reference)喷涨点不足或喷没有对比 reference 中结果。针对一,可找些证据(如列 reference)论证自己方法的涨点幅度和其他 state-of-the-art 的涨点幅度是可比的,「你看,别人发在顶会的结果相比 baseline 也是涨这么多」;针对二,可试着找出这些「证据」方法和自己方法的不同之处或实验细节的不公平之处,比如图像分辨率不同、backbone 不同等;
喷「实验不足」:有条件做实验的,rebuttal 中补上即可;若实验规模太大,rebuttal 期间无条件做出,可在 rebuttal 中承诺 final version 中补上(这样力度会相对较弱);而对于要求不合理的实验意见,可实事求是的说明为何无需做实验;
Rebuttal 时不要漏点,要逐点回应做到有问必答。若因篇幅有限,可将类似的意见合成一点,万不可因篇幅有限擅自删除一些要点或遗漏要点,以免造成含糊不清、浑水摸鱼之嫌,一旦被审稿人发现会在 paper discussion 阶段当作硬伤来「置于死地」;此外,除非基本上全是 positive 评价,要充分利用 rebuttal file 的空间,充分表达自己观点,力图将所有评分均拉到 positive 区间确保十拿九稳;
Rebuttal 时需要揣摩审稿人倾向,「一切可以团结的力量都要团结,不中立的可以争取为中立,反动的也可以分化和利用」。有的审稿人会在意见中明确表示,「如果解决了 xxx,我就会提升评分」,对此一定要充分争取;对于某些审稿人提出的不足(如 novelty),可能刚巧是另一位审稿人提出的优点(「This paper is interesting and novel」),一定要为我所用,让两位审稿人在 paper discussion 中「短兵相接」;对于 borderline 的审稿人,一定要充分「拉拢腐蚀」;对于初审给了 positive 分数的审稿人,一定要巩固基础;对于初审给了 negative 分数的审稿人,一定要放绝大多数的精力和 rebuttal 篇幅来解释澄清,争取「冰释前嫌」;
Rebuttal 是「一盘棋」,整篇 rebuttal 需要统筹协调,与正文、review 配合的相得益彰,同时还需注意 rebuttal 篇幅资源的分配和优化。哪位审稿人应多分配笔墨、哪个问题应多着力回应都需要根据整体审稿意见情况深入思考、统筹安排;
Rebuttal 中能缩写的尽量缩写,如约定俗成的 PCA、CNN 等不必展开,从而节省空间,将资源留给更需要的回应;
Rebuttal 时若发现审稿人的 factual error,如 ta 提出的某个观点有显然错误、提出需要对比的数据集显然不是该领域常用的数据等,作者可在 rebuttal 回应此人时首先指出其错误,先下一城,赢得主动。要知道 rebuttal 除了该审稿人之外,其他审稿人以及 AC 都会看到。此外,这一问题还可以在 AC message(见下文)中指出,降低该审稿人意见在 AC 心中的置信度;
……
Please note that Assigned Reviewer #id has made some statements that are either against the common-sense in our field or self-contradictory (ironically his/her own confidence rating is "very confident"). blabla
We want to bring to your attention the very flawed review #id. This reviewer is self-contradictory, cf. Comment #id1, Comment #id2, and Response #id. blabla
We would like to raise attention to AC that unfortunately Reviewer #id holds a very biased view towards the contributions of our paper. blabla
开头
Thank you for your suggestion.
Thank you for the positive/detailed/constructive comments.
We sincerely thank all reviewers and ACs for their time and efforts. Below please find the responses to some specific comments.
We thank the reviewers for their useful comments. The common questions are first answered, then we clarify questions from every individual review.
We thank the useful suggestions from the reviewers. Some important or common questions are first addressed, followed by answers to individual reviews.
表达同意
We thank the reviewer for pointing out this issue.
We agree with you and have incorporated this suggestion throughout our paper.
We have reflected this comment by …
We can/will add/compare/revise/correct ... in our revised manuscript/our final version.
Due to the rebuttal policy,「authors should not include new experimental results in the rebuttal」, additional results may not be included. However, we will add these mentioned experiments and discussions in our final version. Thank you for the constructive comment.(对于 CVPR/ICCV/ECCV rebuttal 不能提供新结果的政策)
表达不同意
We respectfully disagree with Reviewer #id that ...
The reviewer might have overlooked Table #id ...
We can compare ... but it is not quite related to our work ...
We have to emphasize that ...
The reviewer raises an interesting concern. However, our work ...
Thank you for the comment, but we cannot fully agree with the comment. As stated/emphasized ...
You have raised an important point; however, we believe that ... would be outside the scope of our paper because …
This is a valid assessment of …; however, we believe that ... would be more appropriate because ...
解释澄清
We have indeed stated/included/discussed/compared/reported/clarified/elaborated ... in our original paper ... (cf. Line #id).
As we stated in Line #id, ...
We have rewritten ... to be more in line with your comments. We hope that the edited section clarifies …
额外信息与解释
We have included a new figure/table (cf. Figure/Table #id) to further illustrate…
We have supplemented the xxx section with explanations of ...
Thank you for the comment. We will explore this in future work.
Does rebuttal matter?
Aliaksandr Birukou, Joseph Wakeling, Claudio Bartolini, Fabio Casati, Maurizio Marchese, Katsiaryna Mirylenka, Nardine Osman, Azzurra Ragone, Carles Sierra, and Aalam Wassef. Alternatives to peer review: Novel approaches for research evaluation. Frontiers in Computational Neuroscience, 2011, 5:56.
CVPR/ICCV/ECCV 等会议在使用 Toronto Paper Matching System 作为准则之一进行论文分配:http://www.cs.toronto.edu/~zemel/documents/tpms.pdf
指审稿人或 AC 在投稿池子中申请审稿,按程度不同可大致分为:eager、willing、in a pinch、not willing 等。AAAI/IJCAI 等会议在沿用此方式。
http://cvpr2019.thecvf.com/files/CVPR%202019%20-%20Welcome%20Slides%20Final.pdf
Yang Gao, Steffen Eger, Ilia Kuznetsov, Iryna Gurevych, Yusuke Miyao. Does My Rebuttal Matter? Insights from a Major NLP Conference. NAACL-HLT 2019
关注公众号:拾黑(shiheibook)了解更多
[广告]赞助链接:
四季很好,只要有你,文娱排行榜:https://www.yaopaiming.com/
让资讯触达的更精准有趣:https://www.0xu.cn/
随时掌握互联网精彩
- 1 奋力打开改革发展新天地 7936140
- 2 中国黄金原董事长家搜出大量黄金 7994581
- 3 保时捷断臂求生 7891852
- 4 “冷资源”里的“热经济” 7700036
- 5 赖床其实是在保护心脏 7685246
- 6 刘强东提前发年终奖 7559961
- 7 向佐红毯小牌大耍 7452723
- 8 喝水后有4种表现提示肾有问题 7349934
- 9 十来岁男孩尾随女孩进电梯脱裤子 7212272
- 10 恒大地产1.2亿元债权将1元起拍 7178350