Get your International Conference on Machine Learning (ICML) paper right the first time. The template handles the official structure, margins, and references, so you can focus on the work with confidence.
4.5
Authored by Sumalatha Gangadhar | Published on 25 July, 2026
A ICML conference paper is most often sent back for formatting, not content: margins, front-matter order, and page numbering. The proceedings chairs checks these before anything else. This template sets the correct structure from the first draft, following the International Conference on Machine Learning (ICML) formatting guidance.
It is a formatting shell for a ICML conference paper, built to the official rules. It covers the title page, approval/signature page, abstract, table of contents, chapters, references, and appendices, with a consistent citation style. Confirm the current requirements in the official guidelines before you begin. See also the author resources.
Set up the front matter and chapters in the required order, apply the correct margins and spacing from the manual, and use the prescribed citation style throughout. Keep figures and tables captioned and numbered as specified.
The signature failure mode is margin, front-matter order, and page-numbering rules. Second is inconsistent citations. Third is missing or out-of-order front matter. Fix all three against the official guidelines before submitting.
The shell mirrors what the proceedings chairs checks first: structure order, margins, and citation consistency. Use the International Conference on Machine Learning (ICML) formatting guidance as the source of truth.
| Element | ICML requirement |
|---|---|
| Figures/tables | Captioned and numbered per manual |
| Keywords | Not requested. ICML does not require authors to provide keywords with their submissions. The conference lists topical areas for submissions (e.g., general machine learning, deep learning, reinforcement learning, etc.) but these are not author-supplied keywords. |
| References | APA reference format (author-date style). Use natbib.sty and icml2026.bst for LaTeX submissions. In-text citations include authors' last names and year (e.g., Samuel, 1959). References must include page numbers whenever possible, be placed in chronological order when multiple citations appear together, and be alphabetized by first author surnames. Authors are required to replace arXiv citations with peer-reviewed papers where possible. |
| Peer review | Double-blind (double-anonymous). Reviewers do not know the authors' identities and vice versa. Authors must anonymize submissions and refer to their own prior work in the third person. For accepted papers, the original submission, reviews, meta-reviews, rebuttals, and discussion are published on OpenReview alongside the camera-ready version. |
| Submission | https://openreview.net/group?id=ICML.cc/2026/Conference |
| Structure | Title page, front matter, chapters, references, appendices |
| Margins/format | Per the official manual |
| Front matter | In the prescribed order |
| Citations | Per the official guidelines |
| Source of truth | media.icml.cc |
Read the official manual before drafting and before submission. Set margins and front-matter order first. Apply the citation style from the first reference. Compare against a recently accepted ICML conference paper in your department.
Keep numbering and captions consistent. Verify every required page is present and in order. Do a dedicated formatting pass against the official guidelines before submission.
Do one focused pass for margins, front-matter order, and page numbering. Then confirm citation style and that all required pages/sections are present, following the International Conference on Machine Learning (ICML) formatting guidance.