I was recently thinking of attending IEEE ICDM 2018 (International Conference on Data Mining) in Singapore, next month. It is a top 5 data mining conference. According to my schedule, I could attend it for 2 days, and since Singapore is close to China, it is convenient to go there. However, I was quite surprised by how expensive the registration fee of this conference has became. As of today the “standard registration fee (by 28 October)” is roughly 1360 USD$ or 9300 CNY.
This is actually the most expensive conference that I have ever considered attending. Most conferences that I have attended have been in the 300-700 USD range, twice less than ICDM. But is it an outlier? To see more clearly, I decided to compare the standard registration of ICDM 2018 with those of previous editions of ICDM:
- ICDM 2018: 1360 $ USD (11 % increase from 2017)
- ICDM 2017: $1220 USD (12% increase from 2015)
- ICDM 2015: 1080 $USD (28 % increase from 2013)
- ICDM 2013: 844 $USD (68% increase from 2011)
- ICDM 2011: about 500 $ USD
This is quite interesting. It shows a steady increase in the registration price of the ICDM conference over the years. The registration fee has increased so much, that the price is now 2.7 times higher than 8 years ago!
Why is it so expensive?
One could argue that the reason is the location of the conference. But the increase has been steady over the years no matter where the conference was organized. Moreover, such big conferences have often thousands of attendees, and usually many sponsors. I recently attended the KDD 2018 conference, which was also expensive, but less than ICDM. There was about more than 3000 attendees, and if I remember they received more than 1 million dollars in sponsorship.
Thus, where all this money goes? A good part goes to renting a convention center, publishing the proceedings and other aspects such as providing scholarships to students. But many conferences also make some considerable profit. Some conferences are not for profit, while some other conferences will pay the local organizers or the association organizing the conference. I am not sure about how the money is used in the case of ICDM or IEEE and what they will do with the profits, as I could not find the information. But I believe that such big conferences can generate a huge amount of money. By discussing with organizers of smaller conferences (200 attendees) that have much lower registration fees and less sponsorship, I know that some conferences can still make 20,000$ profit.
About IEEE, it is not their only conference in the 1000$ USD range. Some other flagship conferences like IEEE ICC (about communication) also have fees greater than 1000$ USD. In the field of data mining, the KDD conference is also quite expensive, although still less than ICDM. In some ways, manypeople want to attend these conferences so they are willing to pay these high fees.
Consequences of high registration fees
The consequence of such high registration fees is that some people may not have enough money to attend, and that a lot of money is spent by researchers. And in many case, that money comes from research projects funded by the government. Thus, one could argue that this money could be used in better ways.
Personally, I was thinking of attending ICDM but when I saw that I would have to pay almost 1400 $ USD for two days, I think it is not reasonable to spend that much money. I have enough research funding but I still do not want to waste the money provided by the government for supporting research. Thus, this year, I will use the money for other things rather than going to ICDM.
Philippe Fournier-Viger is a professor, data mining researcher and the founder of the SPMF data mining software, which includes more than 100 algorithms for pattern mining.