The International Conference on Machine Learning (ICML) is an annual academic conference covering the field of Machine learning. It has been held every year since 1980, making it the longest-running venue dedicated to the discipline. ICML is organized by the International Machine Learning Society (IMLS) and is widely considered, along with NeurIPS and ICLR, one of the three primary conferences of highest impact and reputation in machine learning and Artificial intelligence research. Precise dates vary annually, with paper submissions generally due at the end of January and the conference held the following July.
ICML began as the International Workshop on Machine Learning, first held at Carnegie Mellon University in Pittsburgh in July 1980. It was organized by Jaime Carbonell, Ryszard S. Michalski, and Tom M. Mitchell. In 1993 the event transitioned from a workshop series to a full conference under its current name. Since the 2010s, proceedings have been published open-access in the Proceedings of Machine Learning Research (PMLR), which grew out of the JMLR Workshop and Conference Proceedings and was renamed PMLR in 2015.
Growth and Impact
The conference has published numerous foundational papers in modern AI. Landmark contributions presented at ICML include K-means clustering via PCA from Lawrence Berkeley National Laboratory (2004), Google's batch normalization paper, Google Brain's EfficientNet (2019), OpenAI's CLIP (2021), and UC Berkeley's Soft Actor-Critic (2018). These works have shaped deep learning architectures and training methods used across the industry.
Submission volume has grown exponentially over the past decade. The number of submissions rose from 1,037 in 2015 to 12,107 in 2025, with accepted papers increasing from 270 to 3,260 in the same period. Acceptance rates have stayed between roughly 21 percent (2020) and 30 percent (2024). In 2026, the conference set a record with 24,371 submitted papers, more than double the prior year. Of those, about 2.2 percent were selected as spotlight papers and 0.7 percent as oral presentations, while the rest of accepted submissions were presented in poster form. An annual jury selects the best contribution for the Outstanding Paper Award, formerly called the Best Paper Award.
Scope and Focus
ICML covers the full breadth of Machine learning, with special emphasis on theoretical analysis, algorithmic innovation, and statistical learning. Compared to NeurIPS and ICLR, ICML traditionally features more content on statistical learning theory, reinforcement learning, robotics, and optimization theory. This distinct profile attracts researchers working on mathematical foundations as well as practical systems.
Industry presence at ICML is substantial. Technology companies, including Google, Microsoft, Amazon, Meta, and Apple, have regularly appeared among sponsors, using the conference to publish research and recruit talent. Financial firms such as Citadel Securities, Jane Street Capital, and D.E. Shaw also maintain a strong on-site presence. The conference serves as a significant meeting point between academic research and industrial deployment, where advances in areas like Deep learning often get their first large-scale peer review.
Organizational Structure and Relations
ICML is organized under the International Machine Learning Society, which also oversees the IMLS website and supports the conference's long-run institutional continuity. The conference collaborates with affiliated journals, including the Journal of Machine Learning Research, and the edited machine-learning journal, to manage cross-publication and archival references.
Since its founding at Parc's era of academic workshops, ICML has retained a workshop-like sense of collegial exchange despite its scale. With over twenty thousand submissions in recent years, the organizing committee dispatches a formally selected review process and site-wide poster sessions to handle the load. The conference’s publisher, PMLR, provides open access to proceedings, ensuring that published work reaches a global audience free of paywalls. As of 2026, the conference has held a continuous annual record of events across multiple locations in North America and Europe in the rotating format.
See Also
- NeurIPS
- ICLR
- Journal of Machine Learning Research
- Machine Learning journal