ICML 2024 Author Instructions
Paper Submissions
Submitted papers are composed of a main body, which can be up to eight pages long, followed by unlimited pages for references and an appendix, all in a single file. Note that upon the acceptance of the paper, ICML will publish the submission as a whole: That is, the camera ready version will include the appendices (and references). For details concerning the format of the papers, please see the LaTeX style files, an example paper, and the Call For Papers. There is no support for any software other than LaTeX. All submissions must be via OpenReview, anonymized, and must closely follow the formatting guidelines in the templates; otherwise, they will automatically be rejected. In particular, any submission whose main body goes over the 8 page limit will be automatically rejected. (For the final version, an extra page will be allowed. See the example paper for further information.)
As in previous years, authors have the option of uploading extra files to provide further details of their work. The files included in the supplementary may contain code that supports experimental findings, or extra data, or other (anonymized) papers of the authors whose results are needed by the submitted paper. It is entirely up to the reviewers to decide whether they wish to consult any of the appendices in the submitted paper or this additional material. Therefore, if there is material critical to the evaluation of the paper, it needs to be included in the main body of the paper.
Authors are encouraged to submit code to foster reproducibility. Reproducibility of results and easy availability of code will be taken into account in the decision-making process. Authors should avoid submitting links to non-anonymized repositories and instead submit the code base itself or anonymized repositories.
Authors will be asked to confirm that their submissions accord with the ICML code of conduct. We require that each accepted paper have at least one author with a in-person conference registration even if in-person attendance is not possible.
Supplementary Material
ICML 2024 supports the submission of two kinds of supplementary material: supplementary manuscripts and code/data. In particular, if an anonymous reference is made in the paper, authors should upload the referenced papers, so that the reviewers can check the results in the referred paper. The supplementary material must also be anonymized. Note that traditional text appendices to the paper need not be submitted as a separate supplementary material; as mentioned above, unlimited appendices are allowed in the main submission file of a paper.
The supplementary code can be submitted as either a zip file or a pdf. For code submissions, we expect authors to anonymize the submitted code. This means that author names and licenses should be removed. Submission of code through anonymous GitHub repositories is also allowed; however, they have to be on a branch that will not be modified after the submission deadline. Please enter the GitHub link in a standalone text file in a submitted zip file. Data submissions (provided that the authors have the right to do so) in anonymous repositories are welcome.
Supplementary material will not be published or archived, and there are no format restrictions. Authors are therefore responsible for the archival and access of the supplementary if they want to refer to it in the final version of their paper.
There is no separate deadline for supplementary material, and all such material must be submitted by the same deadline as for paper submission.
Double Blind Reviewing
Reviewing for ICML 2024 is double-blind; i.e., reviewers will not know the authors’ identity (and vice versa). Detailed instructions for how to ensure anonymity are also contained in the above example paper. In brief, authors should refer to their prior work in the third person wherever possible. They should refrain from including acknowledgements, grant numbers, or public github repository links in their submissions.
Previously published papers with substantial overlap written by the authors must be cited in such a way so as to preserve author anonymity. Differences relative to these earlier papers must be explained in the text of the submission. For example: “This work builds on [reference], which showed that…”. If an anonymous reference is needed in the paper (e.g., for referring to the authors’ own work that is under review elsewhere), include the referred work as supplementary material as noted above. Note that anonymizing the submissions is mandatory, and papers that explicitly or implicitly reveal the authors’ identities will be rejected.
Reviewers may be able to deduce the authors’ identities by using external resources, such as technical reports published on the web. The availability of information on the web that may allow reviewers to infer the authors’ identities does not constitute a breach of the double-blind submission policy. Reviewers are explicitly asked not to seek this information.
Reviewing and Author Response
Papers will not be public during the review period. Only accepted papers will be made public through OpenReview, and only after the camera-ready deadline. Reviewers are forbidden from sharing papers they receive for review, or using the material in any way other than to provide their review.
After initial reviews, authors will have the opportunity to respond to reviewer comments. During the rebuttal period, authors can see the reviews and respond to their content, but these responses will be visible to the reviewers and ACs after the rebuttal period, when discussion starts. During the subsequent discussion period, authors will be able to discuss with the reviewers and ACs.
Authors can submit multiple responses per submission. Any of the authors of a paper can enter/edit the response, and the response can be returned to and edited up to the deadline for author feedback. As reviewing is double-blind, authors should not include any identifying information in their response. No non-anonymized URLs can be included in the response. Authors should use public cloud services to include information, rather than custom websites or e.g. tinyurl links (which in theory could e.g., track IP information).
We recommend using judgment when crafting a response. There is no need to respond to every minor question or suggestion for improvement. Rather, the response is a good opportunity for addressing issues like a reviewer’s uncertainty about a point, a reviewer making an incorrect assumption, or a reviewer misunderstanding some part of the paper. A professional and polite language is generally the most effective.
We aim to provide three reviews for every paper. The reviewer ids are arbitrary, do not read much into them. The structure of the author response is up to the authors. It is typical to organize the response by reviewers. Here, use the reviewer ids to refer to the particular reviews.
There is no option to upload a revised version of the paper during the author feedback period. Upon the acceptance of the paper, it is up to the authors to include any changes to improve the paper (without essentially changing its content, compared to what the reviewers have seen) in the final, camera ready version of the paper.
Dual Submission Policy
It is not appropriate to submit papers that are substantially similar (or identical) in content to versions that have been previously published, or accepted for publication, or that are under review at other peer reviewed conferences or journals at any time while they are also under consideration for ICML. Such submissions violate our dual submission policy and will be rejected.
Submission is permitted for papers presented or to be presented at conferences or workshops without proceedings (e.g., ICML or NeurIPS workshops), or with only abstracts (limited to at most 4 pages) published.
Submission is permitted for papers that have previously been made available as a technical report or preprint. In this case, we suggest the authors not cite the report to preserve anonymity.