Where to Submit Machine Learning Work
Published · paper counts are current, not as-published
Machine learning runs the most venues of any field indexed here, and it is also the only one where the usual advice actually holds: pick the venue that matches your subject. In the other fields that advice is empty, because their venues publish the same things. Here they do not.
Here the Venues Really Do Differ
Ranking each venue's topics and intersecting the top three across the field leaves nothing at all . Computer vision, by comparison, leaves Multimodal Learning shared by every one of its venues. The leading topic here takes 5 different values across 10 venues - theory, causal work and the general meetings are not competing for the same paper.
| Venue | Papers | Editions | Years | Runs | Leading topic |
|---|---|---|---|---|---|
| NIPS | 21,950 | 20 | 2006-2025 | every year | Optimization |
| AAAI | 20,013 | 8 | 2019-2026 | every year | Representation Learning |
| ICML | 14,281 | 13 | 2013-2025 | every year | Optimization |
| ICLR | 11,015 | 8 | 2018-2025 | every year | Optimization |
| IJCAI | 9,941 | 12 | 2013-2025 | irregularly | Representation Learning |
| AISTATS | 4,802 | 18 | 1999-2025 | irregularly | Optimization |
| COLT | 1,791 | 14 | 2013-2026 | every year | Learning Theory |
| UAI | 1,698 | 8 | 2019-2026 | every year | Causal Inference |
| ALT | 450 | 2 | 2025-2026 | - | Learning Theory |
| PGM | 220 | 1 | 2020-2020 | - | Probabilistic Modeling |
And No Single Venue Owns It
That is the flattest distribution of the three fields with enough venues to compare, and it follows from the first finding rather than being separate from it: when venues specialise, no one of them can absorb the field. The smaller ones here are not overflow for work rejected higher up - they are where particular kinds of work belong.
What This Does Not Show
- Not deadlines, and not acceptance rates. The corpus records published papers, not submission dates or review outcomes.
- Topic labels are assigned, not declared. They come from a classifier, so "leading topic" means the label most often applied to that venue's papers - see the methodology page for which model and what it costs.
- 1 venue left out of the table. It holds papers in this corpus but no year above the threshold that separates an edition from partial ingestion, so there is nothing to put in a calendar.
- A field is not its indexed venues. Machine-learning work also appears at the applied venues and in journals outside this set.
The same comparison for computer vision and natural language processing. The numbers here are read from the database when you load the page.