If you follow AI news, you will often see papers described as “accepted at NeurIPS” or “presented at CVPR.” These names refer to academic research conferences, where scientists share new work after it has been checked by other experts. For researchers, a paper at one of these events can be an important career milestone.
So what actually happens at these meetings, and which one fits your work? Below you will find how they operate, what each major conference is known for, and confirmed dates and locations for the 2026 and 2027 editions.
How research conferences work
Research conferences are different from business events. Their main purpose is to publish and discuss new scientific results rather than to sell products.
Papers and peer review
A paper is a written report on a new method, experiment or theory. Before a conference, researchers submit papers, and other experts in the field review them. This process is called peer review. Reviewers judge whether the work is new, correct and useful, and the conference accepts only part of what it receives.
In AI, these conferences act much like journals do in other sciences. Accepted papers are published in the conference proceedings, which is the official collection of papers from that year.
Talks and posters
Accepted papers are usually presented in one of two ways. An oral presentation is a short talk in front of an audience. A poster presentation uses a large printed board that summarizes the paper, and the authors stand beside it to answer questions. Poster sessions are often the best place to have real conversations, because you can talk directly with the people who did the work.
Workshops and tutorials
Most large research conferences also host workshops and tutorials. A workshop is a smaller meeting on one narrow topic, often with its own papers and invited speakers. A tutorial is a longer teaching session that introduces a subject to people who are new to it. NeurIPS, for example, describes its workshops as a less formal setting for the exchange of ideas.
The major conferences and what they are known for
NeurIPS
NeurIPS (Neural Information Processing Systems) was founded in 1987, and its first meeting was held in Denver, Colorado. Today it describes itself as a yearly, interdisciplinary meeting with many tracks. Its program includes invited talks, demonstrations, oral and poster presentations, tutorials and workshops. It also runs an exhibition focused on machine learning in practice.
ICML
ICML (the International Conference on Machine Learning) describes itself as “the premier gathering of professionals” working to advance machine learning. The 2026 edition in Seoul was the forty third conference. Its attendees range from academic and industry researchers to entrepreneurs, engineers and graduate students.
ICLR
ICLR (the International Conference on Learning Representations) focuses on representation learning, a branch of AI that is often called deep learning. The first ICLR took place in 2013. From the start it used an open review model, in which reviews are published online and anyone can comment on submitted papers.
CVPR, ICCV and ECCV
Each of these three describes itself as a premier event for computer vision, the area of AI that helps machines understand images and video. CVPR (Computer Vision and Pattern Recognition) is an annual event with a main conference and many co-located workshops. ICCV (the International Conference on Computer Vision) is held every two years and is sponsored by the IEEE, a large engineering organization. ECCV (the European Conference on Computer Vision) is also held every two years, in even years, and is managed by the European Computer Vision Association. Because ICCV and ECCV alternate, the computer vision community usually has two major meetings each year.
ACL, EMNLP and NAACL
These conferences cover natural language processing (NLP), which is the study of how computers work with human language. They are linked to the Association for Computational Linguistics (ACL), which was founded in 1962 and took its current name in 1968. ACL is the association’s flagship conference. EMNLP (Empirical Methods in Natural Language Processing) is organized by SIGDAT, a special interest group within ACL. NAACL is the conference of the association’s Nations of the Americas Chapter.
AAAI
The AAAI Conference is run by the Association for the Advancement of Artificial Intelligence, which was founded in 1979. The conference aims to promote research across AI and its related fields, and the association’s stated mission focuses on safe and responsible AI that benefits society.
IJCAI
IJCAI (the International Joint Conference on Artificial Intelligence) has a long history. Its conferences have run since 1969, first every two years and then annually from 2015. Like AAAI, it covers the whole field of AI rather than one narrow topic. In 2026 it joined with ECAI, the European AI conference, for a combined event in Bremen.
KDD
KDD is organized by ACM SIGKDD, the special interest group of the Association for Computing Machinery for knowledge discovery, data science and AI. The conference focuses on data mining, which means finding useful patterns in large datasets. Alongside its research track, it has an applied data science track for work on real world uses.
AISTATS
AISTATS (Artificial Intelligence and Statistics) has met since 1985. It welcomes work at the intersection of AI, machine learning and statistics, and describes itself as an interdisciplinary gathering of researchers.
Who should attend?
Research conferences are most valuable for people who read or write academic papers. That includes university researchers, PhD students and industry scientists. Engineers who want to stay close to the latest methods can also benefit, especially from tutorials and poster sessions.
Recruiters and companies also attend, because these events gather many skilled researchers in one place. However, if your main goal is to buy AI tools or learn basic skills, an industry expo or a practical workshop may suit you better. Our guide on how to choose the right AI event on the blog can help you decide.
Many conferences offer reduced registration for students, and some provide travel grants or volunteer roles. Check each conference’s official website for current options, as these change every year.
Tips for your first research conference
Plan ahead, because these events are large and busy. Read the list of accepted papers before you arrive. Mark the posters you want to visit, since poster halls at large conferences can be very crowded.
Do not try to attend every talk. Pick one or two workshops that match your interests and spend real time there. Workshops are smaller, so it is easier to meet people and ask questions.
Finally, bring a short way to explain your own work. Many good conversations start at a poster and continue over coffee. Our first conference checklist has more practical tips.
Confirmed dates for 2026 and 2027
The table below shows confirmed dates and locations from official conference websites. Some 2026 events have already taken place and are included for reference. Where a 2027 edition is not listed, its dates had not been confirmed when we checked.
| Conference | 2026 edition | 2027 edition |
|---|---|---|
| NeurIPS | Dec 6 to 12, Sydney (satellites in Atlanta and Paris, Dec 9 to 13) | Not yet confirmed |
| ICML | Jul 6 to 11, Seoul | Not yet confirmed |
| ICLR | Apr 23 to 27, Rio de Janeiro | Apr 26 to 30, California (venue to be announced) |
| CVPR | Jun 3 to 7, Denver | Jun 20 to 25, Seattle |
| ICCV / ECCV | ECCV: Sep 8 to 12, Malmö | ICCV: Hong Kong; dates awaiting clarification* |
| ACL | Jul 2 to 7, San Diego | Aug 17 to 22, Kyoto |
| EMNLP / NAACL | EMNLP: Oct 24 to 29, Budapest | NAACL: Jun 1 to 5, San Francisco |
| AAAI | Jan 20 to 27, Singapore | Feb 16 to 23, Montréal |
| IJCAI | Aug 15 to 21, Bremen (with ECAI) | Aug 7 to 13, Kyoto (satellite in Hengqin, Aug 15 to 17) |
| KDD | Aug 9 to 13, Jeju | Aug 1 to 5, San Jose |
| AISTATS | May 2 to 5, Tangier | May 3 to 6, Montréal |
Late in the year, EMNLP 2026 in Budapest (October 24 to 29) and NeurIPS 2026 in Sydney (December 6 to 12) close out the 2026 research calendar. To plan beyond that, open our 2027 event archive or filter by machine learning and computer vision topics.
ICCV 2027 date note, checked 24 October 2026: the official conference page gives October 3–7 in its heading and October 2–8 in its venue text. Hong Kong is confirmed; reconfirm the dates with the organizer before making travel plans.
Key takeaways
-
Research conferences publish peer reviewed papers and are central to how AI research is shared.
-
NeurIPS, ICML and ICLR cover machine learning broadly.
-
CVPR, ICCV and ECCV focus on computer vision, while ACL, EMNLP and NAACL focus on language.
-
AAAI and IJCAI cover the broad field of AI, KDD focuses on data mining, and AISTATS links AI with statistics.
-
Always check official websites for dates, because some 2027 editions are not yet confirmed.
Frequently asked questions
Can I attend without an accepted paper?
Yes. Most research conferences allow anyone to register and attend. Authors of accepted papers must usually register, but many attendees come only to learn and network.
How does a poster differ from an oral presentation?
An oral presentation is a short talk to a seated audience. A poster is a printed summary of the paper, presented during a session where visitors walk around and ask questions. Both formats are used for accepted papers.
Which conference is best for beginners?
These events are technical, so beginners should focus on tutorials and poster sessions, which are easier to follow. Larger conferences such as NeurIPS offer a wide range of tutorials across many topics.
Why do some conferences have satellite locations?
A satellite event is an extra location that runs alongside the main conference. For example, NeurIPS 2026 has satellite events in Atlanta and Paris in addition to the main meeting in Sydney, which can make travel easier for some attendees.
