First International Workshop on Machine Learning held at Carnegie Mellon

Why a dedicated meeting mattered

Machine learning draws on statistics, pattern recognition and artificial intelligence, and researchers working in those traditions do not automatically read the same papers or use the same vocabulary. A regular meeting devoted to learning itself gives them a shared venue: a place to compare methods, argue about how results should be measured, and hear about work in progress before it reaches a journal.

July 1980, Carnegie Mellon

The first International Workshop on Machine Learning was held at Carnegie Mellon University in Pittsburgh in July 1980^[1]. It was organized by Jaime Carbonell, Ryszard S. Michalski and Tom M. Mitchell^[1]. The title said workshop rather than conference, and the word carries meaning: a workshop is typically a smaller working meeting where researchers bring unfinished results and spend time on discussion, rather than a stage for finished papers.

Becoming ICML

In 1993 the meeting turned into a full conference series under its current name, ICML^[1]. The change of name and format reflects a field that had grown: once the number of submissions outgrows informal arrangements, a meeting needs review procedures, published proceedings and a standing organization.

The numbers

Growth is easiest to see in submissions. In 2015 the conference received 1,037 submissions and accepted 270; in 2025 it received 12,107 and accepted 3,260^[1]. Submissions rose roughly twelvefold between those two years, while the share of papers accepted stayed in a similar range, at about a quarter. The expansion of the conference therefore tracks the expansion of the field rather than a loosening of standards.

Where ICML sits

ICML is the oldest of the three primary machine learning conferences, alongside NeurIPS and ICLR^[1]. The series is organized by the International Machine Learning Society (IMLS)^[1]. The conference's own website describes it as the premier gathering of professionals dedicated to the advancement of the branch of artificial intelligence known as machine learning^[2]. Official proceedings appear on the Proceedings of Machine Learning Research site, proceedings.mlr.press^[2]. That arrangement means a paper presented at the meeting stays findable and citable long after the event, and the record of what the field considered worth publishing accumulates in public.

Why it matters

A conference does more than distribute papers. It sets expectations about what counts as evidence, which comparisons are fair, and which problems are worth working on next. The workshop that met at Carnegie Mellon in July 1980 began that process for machine learning, and the sequence of meetings that followed kept it going: each year's accepted papers become the reference point that the next year's submissions have to answer to.