The First ICLR Conference, formally the inaugural edition of the International Conference on Learning Representations, took place in 2013. It was founded by Yann LeCun and Yoshua Bengio in 2012, with the first event held in late April or early May of 2013. The conference was established to provide a dedicated venue for research on learning representations, a core area of Machine learning and Deep learning. Along with NeurIPS and ICML, ICLR has since become one of the three primary conferences of highest impact and reputation in machine learning and Artificial intelligence research.
The founding of ICLR was motivated by a perceived gap in existing conferences for focused discussion on representation learning, which underpins many advances in Neural network models. From its first edition, the conference adopted an open peer review process for refereed papers, a model proposed by Yann LeCun. This process made submissions and reviews publicly visible, a departure from traditional double-blind review systems. The 2013 event featured invited talks as well as oral and poster presentations of accepted papers, a format that has persisted in subsequent editions.
Founding and Organizers
The conference was co-founded by Yann LeCun, then at New York University, and Yoshua Bengio of the University of Montreal, both prominent figures in deep learning research. Their collaboration began in 2012, with the goal of creating a venue that emphasized learning representations over application-specific results. The open review model was a direct response to perceived inefficiencies in conventional peer review, aiming to increase transparency and speed of feedback. The first edition's program committee included researchers from academia and industry, though the full roster is not widely documented.
Open Peer Review Process
A defining feature of the first ICLR was its open peer review system, where paper submissions, reviews, and author responses were posted publicly on the conference website. This model, proposed by LeCun, allowed for community-wide discussion before and during the review period. Unlike traditional conferences where reviews are confidential, ICLR's process encouraged broader input and iterative improvement of papers. The approach was experimental at the time and has since influenced other venues, though it also drew criticism regarding potential bias and workload for reviewers. The 2013 edition served as a testbed for this process, which remains a hallmark of ICLR.
Program and Content
The 2013 program included invited talks from leading researchers and contributed papers presented as orals or posters. Topics covered included unsupervised learning, deep architectures, and representation learning for vision and language. Specific papers from the first edition are not exhaustively cataloged, but the conference attracted submissions from groups at institutions such as University of Toronto, Stanford AI Lab, and MIT CSAIL. The event was held in a single location, with no satellite workshops in its first year, focusing instead on a compact program of high-quality presentations.
Impact and Legacy
The first ICLR established a template that would grow into one of the most selective and influential conferences in machine learning. Its open review process and emphasis on representation learning helped shape subsequent research directions, particularly in deep learning. By 2025, ICLR consistently ranks among the top venues for AI research, with acceptance rates often below 30 percent. The 2013 edition is recognized as a foundational moment for the field, predating the rise of Transformer (architecture) models and Large language model systems that dominate contemporary research. The conference's success also validated the open review model, which has been adopted in modified forms by other venues.
See Also
- ICML
- NeurIPS
- AAAI Conference on Artificial Intelligence
References
- Source facts provided from Wikipedia (CC BY-SA).
External Links
- Official website (not available in this context).