The Proceedings of Machine Learning Research (PMLR) is an open-access scholarly publication series dedicated to publishing peer-reviewed proceedings from leading academic conferences and workshops across machine learning and artificial intelligence. Formerly known as JMLR Workshop and Conference Proceedings (JMLR W&CP), PMLR provides persistent, open-access bibliographic records, downloadable PDFs, and standardized BibTeX metadata for prominent computational venues.
In modern computer science scholarship, conference proceedings serve as the primary archival record for rapid scientific breakthroughs. As machine learning research expanded beyond conventional annual journal cycles, PMLR emerged as an essential non-profit infrastructure ensuring that conference proceedings remain freely accessible, citable, and indexed without author-facing article processing charges (APCs) or commercial subscription paywalls.
What PMLR Is
PMLR is an open-access electronic series that publishes volume-based conference proceedings. Unlike commercial academic publishers that bundle conference proceedings behind institutional paywalls or transfer copyright away from authors, PMLR operates as an academic-led publishing platform hosted at proceedings.mlr.press.
Each volume within PMLR is separately titled and dedicated to an individual conference or specialized workshop. Every published manuscript includes a stable web landing page featuring the abstract, author list, direct links to full-text PDFs, supplementary materials, software repositories, video recordings, and associated review threads on platforms like OpenReview. Papers published through PMLR receive persistent volume and page designations, enabling comprehensive indexing by bibliographic services such as DBLP, Google Scholar, and CrossRef.
Who Runs It and What It Covers
PMLR is managed by an editorial board of active computer science researchers. The current Series Editors are Tegan Emerson, Hoel Kervadec, and Neil Lawrence, with founding editorial leadership originally established alongside Mark Reid. The publication series operates in philosophical and structural alignment with the Journal of Machine Learning Research (JMLR), preserving the community’s commitment to scholar-owned open-access publishing.
The proceedings series was originally established as JMLR Workshop and Conference Proceedings (holding legacy ISSN 1938-7228). A formal restructuring and rebranding to Proceedings of Machine Learning Research was finalized on April 10, 2017. The inaugural proceedings issued directly on the modern PMLR domain was Volume 54, representing the 2017 International Conference on Artificial Intelligence and Statistics (AISTATS).
PMLR publishes peer-reviewed research presented at major international machine learning conferences and high-impact workshops, including:
- ICML: International Conference on Machine Learning (e.g., Volume 235 for ICML 2024)
- AISTATS: International Conference on Artificial Intelligence and Statistics
- COLT: Conference on Learning Theory
- CoRL: Conference on Robot Learning
- ACML: Asian Conference on Machine Learning
- UAI: Conference on Uncertainty in Artificial Intelligence
- MIDL: Medical Imaging with Deep Learning
- CHIL: Conference on Health, Inference, and Learning
How to Publish and Cite in PMLR
Interacting with PMLR involves two distinct workflows: the volume production pipeline managed by conference organizers and proceedings editors, and the bibliographic citation standards used by authors and researchers.
Conference Volume Production Pipeline
Proceedings publication follows the formal PMLR Proceedings Specification (Version 1.3). Volume editors complete an end-to-end static generation workflow:
- Proposal Submission: Conference program chairs submit an official volume proposal detailing event scope, peer-review acceptance rates, program committee composition, and expected paper counts to the series editors.
- Repository Forking: Once approved, the editorial team establishes a dedicated GitHub repository forked from the official PMLR proceedings template, utilizing the Jekyll static site framework.
- BibTeX Compilation: Editors compile a single master BibTeX file (e.g., icml24.bib) containing an initial @Proceedings volume record followed by individual @InProceedings entries for each accepted paper.
- Identifier and Asset Matching: Each entry must feature a standardized citation identifier structured as lastnameYY (e.g., smith24a). The corresponding final camera-ready PDF must match this key identically (e.g., smith24a.pdf), while supplementary appendices follow the lastnameYY-supp.pdf naming convention.
- Extended Metadata Linking: The specification natively supports fields linking external artifacts, including code URLs, presentation video links, and openreview discussion threads.
- Pull Request and Deployment: Volume editors complete a mandatory quality checklist and submit a pull request against the PMLR repository. Automated continuous integration scripts validate BibTeX formatting, character escaping, and file integrity before static deployment.
How to Cite PMLR Volumes
When citing papers published in PMLR within academic manuscripts, researchers should use standard BibTeX conference proceedings entries rather than generic web links. A representative BibTeX entry for an ICML paper published in PMLR is formatted as follows:
@InProceedings{pmlr-v235-author24a,
title = {Sample Machine Learning Paper Title},
author = {Author, First and Author, Second},
booktitle = {Proceedings of the 41st International Conference on Machine Learning},
pages = {1234–1255},
year = {2024},
editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine},
volume = {235},
series = {Proceedings of Machine Learning Research},
publisher = {PMLR}
}
In standard prose or Harvard/APA reference lists, citations specify the author, year, paper title, conference proceedings title, series name (Proceedings of Machine Learning Research), volume number, and page range.
Strengths and Limits
Evaluating the structural characteristics of PMLR helps scholars understand its role within the wider academic publishing ecosystem.
Key Strengths
- Diamond Open Access: PMLR provides completely free access for readers without charging author-facing publication fees or subscription tolls.
- Author Copyright Retention: Authors retain copyright under open licenses (such as Creative Commons CC-BY), permitting wide redistribution and self-archiving on preprint servers.
- Community Governance: Editorial control resides with domain researchers rather than corporate publishing conglomerates, prioritizing scientific integrity and reproducibility.
- Transparent Errata Protocol: Minor typographical errors in metadata or abstracts can be corrected transparently through community-submitted GitHub pull requests in the underlying repository.
Operational Limits
- Conference-Mediated Access: Individual authors cannot submit standalone papers directly to PMLR; submissions are accepted exclusively through participating peer-reviewed conferences and workshops.
- Volunteer-Driven Infrastructure: Volume generation depends on open-source Jekyll templates and volunteer series editors, which can lead to variable turnaround times between conference closing and final online volume indexing.
- No Print Distribution: PMLR operates strictly as an electronic venue; institutions requiring bound physical volumes must arrange independent third-party print-on-demand services.
Alternatives to PMLR
Machine learning and artificial intelligence conferences utilize diverse publication channels. The table below compares PMLR with other dominant proceedings models:
Primary Conferences
Access Model
Author Cost (APC)
Hosting Platform
When preparing manuscripts for venues published by PMLR, authors must ensure alignment with venue-specific criteria, such as drafting an ICML position paper or reviewing publication tracks for conferences like ICLR. Authors should also examine relevant publisher guidelines, such as the Elsevier generative AI authorship policy, and verify cross-disciplinary standards between machine learning vs artificial intelligence paradigms.
Frequently Asked Questions
What does PMLR stand for?
PMLR stands for Proceedings of Machine Learning Research. It is an electronic series publishing peer-reviewed conference and workshop proceedings in machine learning and statistics.
What was PMLR called before 2017?
Prior to April 10, 2017, the series was known as JMLR Workshop and Conference Proceedings (JMLR W&CP) under ISSN 1938-7228. The transition to PMLR began with Volume 54.
How can an author correct a typo in a published PMLR paper?
If an abstract or author name has a minor typographical error, authors can log into GitHub, navigate to the specific conference proceedings repository, locate the corresponding markdown file in the _posts/ directory, and submit a pull request. Substantive content alterations to the underlying paper PDF require written authorization from the conference proceedings editors.
Does PMLR charge fees to publish accepted papers?
No. PMLR operates on a diamond open-access model. There are no article processing charges (APCs) for authors, and papers are freely accessible to the public upon publication.
Sources