ICML 2026 — International Conference on Machine Learning
📅 Monday, 6 July 2026 → Saturday, 11 July 2026
📍
COEX, Seoul, 
South Korea
43rd edition of the International Conference on Machine Learning — the premier annual ML research conference, alongside NeurIPS and ICLR. Held July 6-11, 2026 at COEX Convention & Exhibition Center in Seoul, South Korea. Day 1 (July 6) is Expo/Tutorial day; main conference runs July 7-9; workshops July 10-11. Expected ~9,000+ attendees. Sponsored by the International Machine Learning Society. Paper acceptances span deep learning theory, generative models, RL, safety, applications.
ICML, the International Conference on Machine Learning, is one of the field's three flagship venues alongside NeurIPS and ICLR, and the 2026 edition brings it to Seoul. It is where a large share of the year's foundational machine-learning research is first presented and argued over — peer-reviewed papers, tutorials that bring practitioners up to speed on fast-moving subfields, and topic workshops where new directions are stress-tested before they reach the mainstream.
On the programme
The core is the main-track paper programme, spanning deep-learning theory and architectures, optimization, reinforcement learning, probabilistic and generative modelling, and the trustworthy-ML cluster of fairness, robustness, privacy and interpretability. A day of tutorials precedes the main conference, invited keynotes from leading researchers frame the big questions, and the closing workshops are where the field's frontier conversations actually happen.
Where it sits in today's AI
ICML is a reliable barometer for where machine learning is heading. Recent editions have tracked the shift from simply training larger models toward making them efficient, controllable and agentic: reasoning and tool use, post-training and alignment, the emerging science of evaluation, mechanistic interpretability, and theory catching up to large language and multimodal systems. For researchers, ML engineers and PhD students it is both the place to read the year's most-cited work first and to recruit, collaborate and benchmark against the genuine state of the art. Hosting the 2026 edition in Seoul also reflects the growing weight of East-Asian machine-learning research and industry.
