ICML 2027 Call For Papers

The 44th International Conference on Machine Learning (ICML 2027) will be held in Rio de Janeiro, Brazil, July 31-August 6, as an in-person event. In addition to the main conference sessions, the conference will include tutorials, workshops and an expo.

We invite submissions of papers on original and rigorous research of significant interest to the machine learning community for the main conference proceedings. All papers will be reviewed in a double-blind process and accepted papers will be presented at the conference. Papers must be prepared and submitted as a single file: 8 pages for the main paper, with unlimited pages for references, the impact statement, and appendices. There will be no separate deadline for the submission of supplementary material. The final versions of accepted papers will be allowed one extra page for the main paper.

Note: ICML is planning to make some updates to the submission process (but not the submission format). We will announce them by late November, please stay tuned.

Change in policy: two-stage reviewing process: ICML 2027 will review submissions in two stages. In the first stage, the area chair assigned to each paper decides whether it should proceed to the second stage. Papers that would clearly not be accepted receive a first-stage rejection. Every proposed first-stage reject must be confirmed by a senior area chair, and when in doubt, papers go to full review. Authors of papers that receive a first-stage rejection will be notified by February 23, 2027 with a short justification. All other papers go through the usual review process, and final decisions will be announced on May 3, 2027. This process is similar to the expedited initial review used by many top journals, where an editor or senior reviewer assesses whether a submission is sufficiently competitive to warrant a full round of external review. We recognize that any early-stage review process carries some risk that a good paper could be rejected prematurely. The requirement that every first-stage rejection be reviewed by both an AC and an SAC, together with the principle that papers should proceed to full review whenever there is doubt, is intended to minimize this risk. The goal of this process is to reduce the number of low-quality papers that go through full review.

Change in policy: contribution type in submission/reviewing: ICML 2027 asks all authors to select the Contribution Type that best reflects the primary scientific contribution of their submission. The available types are General (most papers), Theory, Application-driven, Concept and Feasibility, Negative Results, and Datasets, Evaluations and Benchmarks. Contribution types were introduced at NeurIPS 2026. They will be used to help guide reviewing and as one factor in reviewer and Area Chair matching. Authors should select the type that best represents the main contribution of the paper, recognizing that submissions may naturally contain elements of multiple types. Please see the full ICML 2027 Contribution Types Policy and Guidance for detailed descriptions and reviewing expectations.

Change in policy: LLM use in Reviewing: ICML 2027 will use a single policy for LLM use in reviewing, corresponding to Policy B of ICML 2026. In short, reviewers may use privacy-compliant LLMs to help understand the paper and related works and to polish their reviews. Delegating judgement and critique of the paper to LLMs is not allowed, and reviewers remain fully responsible for the content of their reviews. A single policy simplifies reviewer assignment and activities permitted under policy B have broad support even among authors/reviewers that chose policy A at ICML 2026 (see the paper). The findings from ICML 2026 also suggest that effect of policy B on paper outcomes is minimal. Details and precise guidance for the reviewers are given on the LLM policy page.

Change in policy: Use of Generative AI by Authors & Disclosure Policy ICML 2027 requires authors to disclose substantive use of generative AI tools in conducting the research or preparing the submission. This includes uses such as research ideation, mathematical reasoning, experimental design, code generation, data analysis, literature synthesis, and substantial manuscript drafting. Routine editorial assistance, such as spelling, grammar, formatting, and minor clarity edits, does not require disclosure. All submissions must include an AI Disclosure Statement. The statement does not count toward the paper page limit. Authors remain fully responsible for the correctness, originality, and integrity of all submitted content, including any material produced with AI assistance. When generative AI is part of the research methodology itself, its use must also be described in the paper with sufficient detail to understand and reproduce the work. Please see the full ICML 2027 Policy on the Use of Generative AI by Authors for requirements, examples, and guidance.

Change in policy: Submission caps and reciprocal reviewing policy. Similar to ICLR 2027, ICML 2027 introduces submission quotas and strengthens reciprocal reviewing to cope with the growth in submissions. No author may appear on more than 20 submissions. Each submission must designate one of its authors as reciprocal reviewer, and an author can be designated as the reciprocal reviewer for at most 6 submissions. Each author may appear on at most one submission in which no author is an eligible reciprocal reviewer. Authors with 3 or more submissions must also serve on the program committee, as a reviewer, area chair or senior area chair. Submissions that do not meet these requirements will be desk rejected. Eligibility criteria and further details are given in the guide on Submission Quotas and Reciprocal Reviewing.

Please review Author Instructions, an Example Paper, and use the correct Style Files. Submitted papers that do not conform to these policies will be rejected without review.

Topics of interest include (but are not limited to):

general machine learning (active learning, clustering, online learning, ranking, supervised, semi- and self-supervised learning, time series analysis, etc.)
deep learning (architectures, generative models, theory, etc.)
evaluation (methodology, meta studies, replicability and validity, human-in-the-loop, etc.)
theory of machine learning (statistical learning theory, bandits, game theory, decision theory, etc.)
machine learning systems (improved implementation and scalability, hardware, libraries, distributed methods, etc.)
optimization (convex and non-convex optimization, matrix/tensor methods, stochastic, online, non-smooth, composite, etc.)
probabilistic methods (Bayesian methods, graphical models, Monte Carlo methods, etc.)
reinforcement learning (decision and control, planning, hierarchical RL, robotics, etc.)
trustworthy machine learning (reliability, causality, fairness, interpretability, privacy, robustness, safety, etc.)
application-driven machine learning (innovative techniques, problems, and datasets that are of interest to the machine learning community and driven by the needs of end-users in applications such as healthcare, physical sciences, biosciences, social sciences, sustainability, and climate etc.)

Similar to last year, we also invite submissions of position papers (call for paper is coming soon). Submissions will be handled separately from the main track submissions.

Papers published at ICML are indexed in the Proceedings of Machine Learning Research through the Journal of Machine Learning Research.

We are recruiting reviewers: if you would like to volonteer, please fill this form: Self-nomination form for reviewing


Important Dates and Submission Site

Submission site opens: January 8, 2027.
Suggested OpenReview account creation deadline: January 8, 2027. (If you do not already have an OpenReview account, please register by this date, otherwise we cannot guarantee that your account will be activated in time.**)
Abstract submission deadline: January 14, 2027 11:59 PM AOE or 08 weeks 05 days 18:31:15
Full paper submission deadline: January 22, 2027 11:59 PM AOE or 09 weeks 06 days 18:31:15

Abstracts and papers can be submitted through OpenReview: https://openreview.net/group?id=ICML.cc/2027/Conference

Note: Position papers should be submitted through a separate OpenReview site, as outlined in the upcoming Call for Position Papers (to be available soon).

**OpenReview: All authors must have an OpenReview account. It is strongly recommended that you sign up for OpenReview (or associate your existing account) with an institutional email. If you sign up for OpenReview with an institutional email, your account will be activated immediately; otherwise it can take up to two weeks for your account to be activated.


Policies
Optional in-Person Attendance

Attendance for authors of accepted papers is optional. After acceptance notifications, the authors will be able to decide by a specified date whether they wish to present their paper in person at the conference or they just wish to include their paper in the proceedings (without presentation at the conference). Regardless of this choice, all the accepted papers will receive equivalent treatment in the proceedings. They will all be eligible for ICML awards as well as for the designations of distinction corresponding to the past “oral presentations” and “spotlight posters.” For proceedings-only papers, at least one of the authors must obtain virtual registration.
Deadlines

Abstract and paper submission deadlines are strict. In no circumstances will extensions be given.
Changes to Abstract/Title/Authorship

Authors should include a full title for their paper, complete author list, and complete abstract in the submission form by the abstract submission deadline. While it is possible to edit the title and abstract until the full paper submission deadline, submissions with “placeholder” abstracts that are substantially rewritten for the full submission risk being removed without consideration. The author list cannot be changed after the abstract deadline. After the abstract deadline, authors may be reordered, but not added or removed.
Double-Blind Review

All submissions must be anonymized and may not contain any information with the intention or consequence of violating the double-blind reviewing policy. Authors are allowed to post versions of their work on preprint servers such as arXiv. They are also allowed to give talks on the work(s) submitted to ICML during the review. However, under no circumstances should the work be advertised as an ICML submission at any time during the review period, i.e., from the time the authors submit the paper to the communication of the accept/reject decisions. If the authors have posted or plan to post a non-anonymized version of the paper online before the ICML decisions are made, the submitted version must not refer to the non-anonymized version.
Dual and Concurrent Submission

Authors may not submit papers that are identical, or substantially similar, to versions that have been previously published, accepted for publication, or submitted in parallel to other conferences or journals. Such submissions violate our dual submission policy, and the organizers have the right to reject such submissions, or to remove them from the proceedings. Any concurrent ICML submissions with an overlapping set of authors will also be treated as prior work (so, for example, if publishing one would render the other too incremental, then this may be considered grounds for rejection). Note that submissions that have been or are being presented at workshops without published proceedings do not violate the dual-submission policy.
Reviewing Criteria

Submissions should report original and rigorous research of significant interest to the machine learning community. All claims must be clearly stated and supported by reproducible experiments and/or sound theoretical analysis. The contributions must be situated in the context of the broader scientific and machine learning research literature, acknowledging and differentiating from relevant prior works as appropriate. See Reviewer Instructions for more information.
Ethical Conduct for Peer Review

Authors are expected to follow standard ethical conduct for peer review. In particular:

Plagiarism in any form is forbidden.
Prompt injection is forbidden.
Advertising work (e.g., in a talk, on social media) as being under submission to ICML during the review period is forbidden.
Any form of collusion, whether explicit or tacit (e.g., where authors cooperate with reviewers, ACs, or SACs to obtain favorable or unfavorable reviews) is forbidden.

If you believe someone may be engaging in unethical conduct, please notify ICML by filling out the Ethics Violation Reporting form. You can also report suspected violations to program-chairs@icml.cc.

All suspected unethical behaviors will be investigated by program chairs, integrity chair, or the ICML Oversight Committee, and individuals found violating the rules may face sanctions and/or have their submissions rejected. We will also collect names of individuals who are found to have violated ethics standards; if individuals representing conferences, journals, or other organizations request this list for decision-making purposes, we may make this information available to them. See Peer-review Ethics for additional details.
Impact Statements

Authors are required to include a statement of the potential broader impact of their work, including its ethical aspects and future societal consequences. This statement should be in a separate section at the end of the paper (co-located with Acknowledgements, before References), and does not count toward the paper page limit. In many cases, where the ethical impacts and expected societal implications are those that are well established when advancing the field of machine learning, substantial discussion is not required, and a simple statement such as:

“This paper presents work whose goal is to advance the field of machine learning. There are many potential societal consequences of our work, none of which we feel must be specifically highlighted here.”

The above statement can be used verbatim in such cases, but we encourage authors to think about whether there is content that does warrant further discussion, as this statement will be apparent if the paper is later flagged for ethics review.

In particular, Reviewers and ACs may flag submissions for ethics review. Flagged submissions will be sent to an ethics review committee for comments. Ethics reviewers do not have the authority to reject papers, but in extreme cases, papers may be rejected by the program chairs on ethical grounds, regardless of scientific quality or contribution.

For tips on writing the impact statement, see this post written for NeurIPS 2020.
Lay Summaries

Authors of accepted papers will be required to submit a short “lay summary” of their work (also called "plain language summary") in OpenReview when submitting their camera-ready paper. As machine learning is becoming a more prevalent topic of interest in society, it is important to participate in science communication with the public. See this paper for guidelines. Additional details and examples will be provided prior to the camera-ready deadline.
Self-Rankings

Continuing the practice introduced at ICML 2026, authors with multiple submissions will be required to provide a self-assessment of the quality of their own submissions through a ranking. Disagreements between author rankings and reviewer scores will be used to flag papers that may warrant additional attention from area chairs and to help guide the assignment of emergency reviewers. Submitted self-rankings will be accessible only to the program chairs, associate program chairs, and integrity chairs. Area chairs will have access only to a coarse measure of disagreement between author rankings and reviewer scores.