This question is related to the 2019 NeurIPS conference attendance question, however ICML and CVPR occur together in June, so we split them into their own question instead of delaying resolution until NeurIPS in December.
As reported by the AI Index (p. 26 of the 2018 report), the number of attendees at major machine learning conferences has grown exponentially since the start of the ML revolution in 2011.
As of 2018, the three largest conferences are Neural Information-Processing Systems (NeurIPS), International Conference on Machine Learning (ICML) and Computer Vision and Pattern Recognition (CVPR).
NeurIPS 2019 takes place in Vancouver in December. ICML 2019 and CVPR 2019 take place back-to-back in Long Beach, CA, in June.
Now we ask:
What will be the average (arithmetic mean) number of attendees in 2019 across ICML and CVPR?
Resolution
Resolution will be determined by an attempt to get attendee estimates directly from conference organisers (as done by the AI Index), or by a credible report of such an estimate from the organisers.
Note also that 1) even if NeurIPS, ICML or CVPR cease to be the three biggest conferences they will still be the inputs to this question; 2) the data is based on estimates from the conference organisers, and not actual counts (hence why NeurIPS 2018 data is exactly "8000"); 3) due to constraints imposed by things like the size of venues, tail-end attendance figures might be a poor proxy for demand.
Data
Underlying data on conference attendees going back to the 1980’s can be found in this spreadsheet.
Most recent values have been:
2018: 5767
2017: 3703
2016: 3355