Call for Papers: ICLR 2025

We invite submissions to the 13th International Conference on Learning Representations (ICLR 2025), welcoming papers from all areas of machine learning. The conference will be held at the Singapore EXPO from April 24th to April 28th, 2025.

For any information not listed below, please submit questions via Contact ICLR Program Chairs or email program-chairs@iclr.cc.

Key Dates (All times UTC-12h, "Anywhere on Earth")

• Abstract Submission: September 27, 2024
• Full Paper Submission: October 1, 2024
• Reviews Released: November 12
• Author/Reviewer Discussion: November 12-26
• Author Last Day to Reply: November 27
• Final Decisions: January 22, 2025
• Camera-Ready Deadline: March 14, 2025

Subject Areas

ICLR considers a broad range of subject areas including:

• Unsupervised, self-supervised, semi-supervised, and supervised representation learning
• Transfer learning, meta learning, and lifelong learning
• Reinforcement learning
• Representation learning for computer vision, audio, language, and other modalities
• Metric learning, kernel learning, and sparse coding
• Probabilistic methods (Bayesian methods, variational inference, sampling, UQ, etc.)
• Generative models
• Causal reasoning
• Optimization
• Learning theory
• Learning on graphs and other geometries & topologies
• Societal considerations including fairness, safety, privacy
• Visualization or interpretation of learned representations
• Datasets and benchmarks
• Infrastructure, software libraries, hardware, etc.
• Neurosymbolic & hybrid AI systems (physics-informed, logic & formal reasoning, etc.)
• Applications to robotics, autonomy, planning
• Applications to neuroscience & cognitive science
• Applications to physical sciences (physics, chemistry, biology, etc.)
• General machine learning

Reviewing Process

• Double-blind reviewing: Reviewers cannot see author names, and authors cannot see reviewer names. Papers on arXiv are allowed.
• Submissions are hosted on OpenReview allowing for public discussions by logged-in users.
• Authors can revise papers up to the submission deadline and during the discussion phase. Significant changes during discussion may be ignored.
• All submitted papers (accepted or rejected) will be de-anonymized after notification, with reviews and comments made public.

Paper Length

• Main text must be between 6 and 10 pages (inclusive).
• This limit is strictly enforced for both initial and final camera-ready versions.
• References do not count towards the page limit, and unlimited additional pages are allowed for the bibliography.
• Appendices are allowed but reviewers are not required to read them.

Style Files and Templates

LaTeX style files are provided at: https://github.com/ICLR/Master-Template/raw/master/iclr2025.zip

Reciprocal Reviewing Requirement

• Authors submitting 3 or more papers must serve as a reviewer for at least 6 papers.
• At least one author of each submission must be registered to review at least 3 papers.
• Authors are exempt if they are serving as an AC, SAC, or organizing chair.
• All authors will be notified to register as reviewers after the abstract deadline. Failure to comply may result in desk rejection.

Code of Conduct and Ethics

• All participants must adhere to the ICLR Code of Conduct and Code of Ethics.
• Authors must read, adhere to, and acknowledge the Code of Ethics during submission.

Dual Submission Policy

• Submissions identical or substantially similar to previously published or submitted work are not allowed.
• Papers on arXiv or presented at non-peer-reviewed workshops do not violate the policy.
• Submissions to ICLR can be modified on arXiv during the review period.

Use of Large Language Models (LLMs)

• LLMs are allowed as general-purpose assist tools.
• Authors and reviewers are responsible for LLM-generated content, including plagiarism or fabrication.
• LLMs are not eligible for authorship.

Withdrawal Policy

• Authors can withdraw papers until notification.
• Withdrawn papers remain on OpenReview in a publicly visible "withdrawn papers" section and will be de-anonymized immediately.