Author Instructions for Full Papers

Paper submission

We invite participants to submit in PDF format a 5-page full paper (included in proceedings) in the area of Deep Learning. References and an appendix can go beyond the page limit.

Authors  can download the latest packaged template here or follow the instructions in the repository to clone the template.  Detailed style guidelines can be found within the main.tex or in the compiled versions.

Double Blind Reviewing

The reviewing process for NLDL is double-blind, meaning that reviewers will not know the authors' identities and vice versa. To ensure anonymity, authors should refer to their prior work in the third person, avoid including acknowledgements, grant numbers, or public repository links in their submissions. 

Previously published papers with substantial overlap must be cited in a way that preserves author anonymity, and any differences relative to these papers should be explained in the submission. Anonymizing the submissions is mandatory, and papers that reveal the authors' identities will be rejected. Reviewers are not prohibited from using external resources to potentially infer the authors' identities, but they are explicitly asked not to seek this information.

Reviewing and Author Response

During the review period, papers will not be public. Only accepted papers will be made public through OpenReview after the camera-ready deadline. Reviewers are not allowed to share the papers they receive for review or use the material for any purpose other than providing their review. After the initial reviews, authors will have the opportunity to respond to reviewer comments and to submit a revised paper. Authors can see the reviews and respond during the rebuttal period. The reviewers and ACs are not forced to reply during rebuttal period, but we encourage them to interact. Authors can submit multiple responses per submission, and any author can enter/edit the response until the deadline for author feedback. There will be a subsequent discussion period where reviewers and ACs will discuss the submission. 

As the reviewing process is double-blind, authors should not include identifying information or non-anonymized URLs in their responses. Authors are advised to use public cloud services for including information. When crafting a response, it is recommended to use judgment and focus on addressing important issues rather than every minor question or suggestion. A professional and polite language is recommended. At least, three reviews will be provided for every paper, and the structure of the author response is flexible. It is common to organize the response by reviewers' IDs. 

There is an option to upload a revised version of the paper during the author feedback period. The AC could request further changes to improve the paper and should be included in the final camera-ready version after acceptance.

Dual Submission Policy

Full paper submissions to NLDL should not be substantially similar or identical to versions previously published, accepted for publication, or under review at other conferences or journals, as this violates the dual submission policy and will result in rejection. However, papers presented or to be presented at conferences or workshops without proceedings, or with only abstracts published, are permitted for submission within certain guidelines. Additionally, submission is allowed for papers previously made available as technical reports or preprints, but in such cases, authors are advised not to cite the report to maintain anonymity.

FAQ

About submission process

• Q. Can we please have an extension on the paper registration or submission deadline?
  A. NO. Furthermore, any incomplete submission or a submission not meeting required criteria will be deleted.

• Q. Can I update my paper’s information (e.g., title, abstract, author list) after the paper registration deadline?
  A. You can update the title and abstract until the paper submission deadline. You can also reorder the author list until the paper submission deadline. However, after the paper registration deadline, you can no longer create new paper submissions or add/delete authors of your submission(s).

• Q. Can I add/remove authors after my paper has been accepted?
  A. NO. After the paper registration deadline, the author list is considered final. Changes to the authorship order may be done freely between the registration and submission deadlines. Further changes to the authorship order following acceptance may be considered, but only in special circumstances.

• Q. Do my co-authors need to create an OpenReview account?
  A. Yes. Before the full paper deadline, every co-author needs to create (or update) an OpenReview profile. The information entered in the profile is used for conflict resolution.

About submission format and contents

• Q. Do the 5-pages include the reference and appendix?
  A. NO, the 5-pages is only for the contents of the paper.  References and the appendix can go beyond the 5-pages.  Note that the reviewers are not expected to read the appendix to evaluate the main paper.  Thus, the main results and claims must be within the 5-page limit.

• Q. Are acknowledgements permitted in the submitted paper?
  A. NO, acknowledgements should only be included in the camera-ready paper.

• Q. Can I link to an external webpage from my submission?
  A. This is strongly discouraged because it runs a high risk of violating anonymity, or circumventing length or deadline restrictions. If you feel you absolutely must link to external materials, see the next question.

• Q. Can I link to additional image or video material from the supplementary material?
  A. Only if absolutely necessary and as long as the double-blind review process and deadline integrity are preserved. To that end, authors need to ensure the following conditions: (1) The hosting site and the linked material does not reveal the identity and affiliation of the authors. (2) The hosting site or apps do not track or identify who viewed the materials. (3) The authors provide a smaller-sized version of their image or video material in the submission's appendix.
  
  Authors bear the responsibility and are advised to proceed with caution not to break the double-blind review process. Note, not all hosting services are available in all regions. Authors should also note that, just like for the supplementary material itself, reviewers are under no obligation to review such additional image or video material.

• Q. Is there a formal length restriction on the abstract?
  A. No, but the abstract should concisely summarize the contents of the paper. While there is no fixed length restriction for the abstract, it is recommended to limit your abstract to approximately 150 words.

• Q. I need to insert the number of the paper in the template.  How can I get it?
  A. You should create a submission on OpenReview.  There you will see the number assigned to your paper.

• Q. I found that my submission is not anonymized correctly.  What should I do?
  A. Contact the Area Chair and the Program Chairs immediately.  If we are already during the review process, the paper will be desk rejected due to the violation of the double blind policy.  If the review hasn't started, the PCs will evaluate case by case the violation.

About the review process

• Q. Can I submit a revised version of the paper during the rebuttal?
  A. Yes.  You can upload a revised version of the paper and update it as needed.  The reviewers will see the final version after the rebuttal period is over.

• Q. Is there an author-reviewer discussion period?
  A. No.  During the rebuttal the comments will be open, but the reviewers are not enforced to interact.  The authors can ask questions and more explanations about the reviews, but shouldn't expect to see a reply from the reviewers.

• Q. Can I write more than one rebuttal post per review?
  A. No.  Formally, you can submit only one rebuttal post per review.  While the comments will be open, we discourage you to write several comments as rebuttals.  In our experience, shorter and to the point communication with the reviewers and area chairs is more efficient.
  Note that you are allowed to submit a revised version of the paper.  In this case, you can point to the changes made in the paper and keep the rebuttal short.

About LLMs

• Q. What is the LLM Policy for authors?
  A. Authors may use any tools they find productive in preparing a paper, but must be aware that they are responsible for any misrepresentation, factual inaccuracy or plagiarism in their paper. Papers containing citations of non-existent material will be rejected when found, and may be rejected without review. Similarly, papers containing obvious factual inaccuracies will be rejected when found and may be rejected without review. It is not a defense to a charge of plagiarism or of inaccuracy to argue that “an LLM did it”. You are responsible for what you submit.

• Q. How will the LLM policy be implemented?
  A. Referees who find inaccuracies should act as they usually would; as should area chairs. Glaring examples of citations to non-existent material can be desk-rejected.

• Q. What is the LLM Policy for reviewers?
  A. Reviewers may use any device, including an LLM, to polish their review wording, but must vouch for, and be responsible for, the accuracy of the review. It is a significant act of reviewer misconduct to allow an LLM to see a submission. PCs interpret showing a submission to an LLM as a deliberate reviewer violation of confidentiality. The PCs reserve the right to report reviewer misconduct to future machine learning and related conferences. These conferences then may take actions, e.g., there was a recent PAMI-TC vote that CVPR reviewer misconduct may lead to a 2-year submission ban.

• Q. What is the reasoning behind the LLM policy?
  A. The action that most likely affects the credibility of the conference is using an LLM to write the summary of the paper. The summary is a necessary part of reviewing, because it compels the reviewer to show what they think the paper is about, and so validates the review. A summary that has been polished by an LLM which hasn’t seen the paper is acceptable, as long as the referee vouches for the review. A summary that has been prepared by an LLM—so one where the LLM sees the paper—is profoundly damaging, because it may allow a review to be prepared without the referee trying to understand the paper.

• Q. How will the LLM policy be implemented?
  A. An author may complain to their AC that a summary (and/or other parts of the review) have been prepared by an LLM that has seen the paper. Such a complaint would need to be supported by an example summary (or other part of the review) prepared by the author giving the paper to an LLM. If this matches the reviewer’s comments sufficiently, ACs will pass the complaint on to PCs who are then entitled, but not required, to act. Complaints must not appear on the rebuttal, but be submitted on a separate form. PCs strongly discourage frivolous complaints. Authors should be aware that a complaint to an AC about a review prepared by an LLM without reasonable evidence in support of that complaint, is wasting the ACs time.