# ICLR Founding (2013)

The International Conference on Learning Representations (ICLR) is a top-tier machine learning conference founded in 2012 by Yann LeCun and Yoshua Bengio, first held in 2013, known for its open peer review process and high impact in AI research.

The International Conference on Learning Representations (ICLR) is a premier academic conference in [machine-learning](https://www.wikiprompt.org/wiki/machine-learning) and [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence), established to focus specifically on the representation learning aspects of deep neural networks. Inaugurated in 2013, ICLR is consistently ranked alongside NeurIPS and ICML as one of the three most influential conferences in the field, attracting thousands of submissions annually. Its distinctive hallmark is an open peer review process, where paper submissions and reviewer comments are publicly visible, a model inspired by proposals from [yann-lecun](https://www.wikiprompt.org/wiki/yann-lecun) and formalized by its co-founders [yann-lecun](https://www.wikiprompt.org/wiki/yann-lecun) and [yoshua-bengio](https://www.wikiprompt.org/wiki/yoshua-bengio) in 2012.

## Origins and Founding

The founding of ICLR arose from a perceived gap in existing conferences, which often treated representation learning as a sub-topic rather than a central research theme. In 2012, Yann LeCun, then at New York University, and Yoshua Bengio of the University of Montreal proposed a new venue dedicated to learning representations, including architectures like [neural-network](https://www.wikiprompt.org/wiki/neural-network)s and [deep-learning](https://www.wikiprompt.org/wiki/deep-learning) models. They drew on earlier experiments with open reviewing, such as those used by the Journal of Machine Learning Research, to design a process that would increase transparency and accelerate scientific discourse. The first edition was held in 2013 in Scottsdale, Arizona, United States, from May 2 to May 4, and attracted over 400 attendees.

The open review format, adopted from the outset, departed from traditional double-blind reviewing. Instead, submissions are posted on a public platform where any registered member can comment, and designated reviewers provide formal assessments alongside the open discussion. This approach has been credited with improving paper quality and fostering community engagement, although it has also faced criticisms regarding reviewer workload and potential biases.

## Growth and Impact

From its first year with approximately 200 submissions, ICLR grew rapidly in scale and prestige. By 2020, the conference received over 2,500 submissions, and by 2024, that number surpassed 7,000, with an acceptance rate below 30 percent for the main track. The conference is typically held in late April or early May, rotating between North American and European cities, with virtual components added after 2020. Its proceedings are published through the OpenReview platform, which continues to host the public review history for every paper.

ICLR's impact is evidenced by its citation metrics and the prevalence of its published papers in advancing [generative-ai](https://www.wikiprompt.org/wiki/generative-ai) and [transformer](https://www.wikiprompt.org/wiki/transformer) architectures. Many foundational works on [large-language-model](https://www.wikiprompt.org/wiki/large-language-model)s, including studies on scaling laws and attention mechanisms, have debuted at ICLR. The conference also introduced notable innovations like the "Reproducibility Challenge," which encourages independent verification of published results, and has established awards for best papers and outstanding student papers.

## Conference Structure

Each ICLR edition spans several days and includes invited talks from leading researchers, oral presentations of selected papers, and extensive poster sessions. The program committee, composed of senior area chairs and reviewers from academia and industry, operates under a dual-track system: a standard submission track and a more speculative "Brainstorm" track introduced in 2020 to encourage unconventional ideas. Notably, ICLR has also experimented with a "Test of Time" award, recognizing papers from a decade prior that have had sustained influence.

Key figures in the conference's history include its founding general chairs, LeCun and Bengio, as well as subsequent program chairs such as [yoshua-bengio](https://www.wikiprompt.org/wiki/yoshua-bengio) (2016) and [koray-kavukcuoglu](https://www.wikiprompt.org/wiki/koray-kavukcuoglu) (2018). The conference maintains strong ties to industry laboratories, with regular contributions from researchers at [openai](https://www.wikiprompt.org/wiki/openai), [google-deepmind](https://www.wikiprompt.org/wiki/google-deepmind), and [anthropic](https://www.wikiprompt.org/wiki/anthropic), reflecting the applied nature of modern deep learning research.

## Criticisms and Evolution

The open review model, while revolutionary, has drawn scrutiny. Some researchers argue that public comments can discourage risky work or create adversarial dynamics. In response, ICLR has periodically adjusted policies, such as allowing authors to opt out of public comments during an initial review period and introducing a two-phase review cycle in 2022 to separate discussion from final decisions. Additionally, the conference's growth has led to concerns about reviewing quality; to address this, ICLR increased the number of area chairs and implemented a "consistency check" system.

Despite these challenges, ICLR remains a central venue for disseminating breakthroughs in [deep-learning](https://www.wikiprompt.org/wiki/deep-learning) and [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence), shaping both academic research and industrial deployment. Its founding principles of openness and focus on representation learning have influenced other conferences, though none have fully replicated its model.

## See Also

- ICML - International Conference on Machine Learning
- NeurIPS - Neural Information Processing Systems
- AAAI Conference on Artificial Intelligence - A broader AI conference with a longer history

## References

- Yann LeCun and Yoshua Bengio, founding statement of ICLR, 2012.
- ICLR Proceedings, OpenReview, 2013-2024.
- Conference statistics and acceptance rates, as reported in official announcements and archival records.

---
Source: https://www.wikiprompt.org/wiki/iclr-founding
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-09-07T21:32:38.861665+00:00