Wednesday, December 05, 2018

NeurIPS 2018 Post

Today was my first day at a NeurIPS conference.  Advertising note, before my thoughts on the experience:
Tomorrow I'll be stationed at a poster 10:45 AM -- 12:45 PM @ Room 517 AB #169
A Model for Learned Bloom Filters and Optimizing by Sandwiching
and at the same time Diana Cai will be stationed at a poster I'm involved with:
Room 210 #26
A Bayesian Nonparametric View on Count-Min Sketch
Please stop by to talk to me if you're around -- Room 517 seems to be a bit off the beaten path and I'm concerned I'll be lonely with nobody to talk to.  And of course stop by to see Diana too.

NeurIPS is just huge.  It's like an extremely large academic conference (of 1000-2000 people) glued together to an extremely large industry conference (of more than that).   Just the academic part (talks and poster sessions, 3 parallel sessions for talks) is a lot, and then there's a trade show with booths from 50+ companies there too.

I think I'd like the academic part more if I was in the area -- I'm coming in from the algorithms perspective, and there's a bit of language gap.  It seems to me that conference suffers from some of the standard aspects of large conferences -- with scope and size that big, you have to look and find the things that are interesting and important to you, because a fair bit probably isn't.  And while I can't tell entirely myself, I'm told by others that, given the size, there's not a lot of "junk" -- the work seems good-on-average.  Also, given the conference size, it seems well organized -- staff managing the people flow, they keep things on time, plenty of room in the poster session (with appropriate food and drink stuff).  I can appreciate the work that must go into to making something this big work.

Because the conference is so big, I've run into a good number of "theory people" here.  As a percentage of attendees, we're probably small, but it's a good number because it's so big.  I kept running into people at the poster sessions, which was nice.

The trade show part is impressive in its own way.  If you didn't know AI was big, this would tell you.  All the big players are there, but there are at least a dozen machine-learning focused companies I haven't heard of, and that's not including the dozen or more consulting/Wall Street/hedge fund firms that are big enough into AI that they want to have a presence here.  I understand there's lots of networking and pre-interviewing and interviewing going on.  On the positive side, I feel like if I wanted to leave academia, I could land a job with a variety of AI-based companies, beyond the big companies.   

Hopefully tomorrow will be interesting as well.


No comments: