By rawpixel on Unsplash

The Good, The Bad, The Weird – 2018 SEG Meeting

The SEG Annual Meeting 2018 in Anaheim, California has concluded. This was my first SEG and I can only compare it to the EAGE Exhibition and Conference I attended. There was light. There was shade. There were oddities.

The Light

Let’s talk about the light. You may have noticed that I am fully riding the machine learning gravy train. Turns out 140 others were accepted to do so at SEG as well. Many interesting applications were shown, some even in 4D seismic, my Phd topic, which is notoriously hard. It was lovely and many talks showed some nice results. Nothing ground-breaking, but things are progressing in a nice direction.

The main application, without a doubt was seismic interpretation in its many forms. My presentation was on facies classification too, no surprise there. Well logs got their fair share of the ML treatment. K-means got sold as “the clustering algorithm” here and there and self-organizing maps are all over the place. Geoscientists really like maps. However, here and there, you would see some glimpses of people digging deep and keeping up with the fastpaced world of ML. DBSCAN was used in some instances and showed some good results. And, I got surprised with a spatiotemporal implementation of Densenets, an application that was initially introduced for traffic prediction. Shell showed some uses of scikit-learn that looked like they are safe enough for production.

In 4D time-warping got its fair share, as time-shift analysis is still a poorly solved problem as of 2018. Several presentations worked on pre-stack or at least angle-stack applications, which seems to reflect the trajectory of the industry.

The SEG created the Digital Arena on the exhibition floor, to somewhat cluster the compute resource vendors. This is where they placed the Repro Zoo experiment. A place, where anyone could drop in and reproduce seminal papers in geoscience. You can find the summary by Matt Hall over at Agile Scientific’s Blog.

In the digital arena, a colourful booth made an entrance as Google pushes for the O&G sector. The booth was well-visited, not just because of the cookies. The best afternoon drinks were had at Spectrum,
This makes a natural transition to the shade. The exhibition was rather empty. Rumour has it that some companies cancelled their booths and despite all effort, it seemed smaller than the EAGE, which surprised me. During the icebreaker and presentations, it also seemed like fewer people were present. But maybe my choice of the session was more obscure.

The Shade

I signed up for the Google and Friends hackathon. However, my presentation slot was early in the morning of the same day and I was taught that it’s good etiquette to attend your own session, so I did. I visited the hackathon later and it did not seem particularly interesting to me. For one, every hackathon I attended has made an effort to create a somewhat inspiring ambience. This one was in a conference room in the conference centre and had a very sterile feel to it. In a short chat I had with some people I knew there, they mentioned that the hackathon wasn’t so friendly after all. The question of how to convert SEGY to JPEG was denied. The reason for that kind of information being “confidential” and “proprietary” by the supervising Google employees. This is when I decided to leave the premise to get lunch and return refreshed to some other sessions that would yield a better ROI. I was surprised by that statement as Google does usually try to give off a more open vibe. My solution for a “patch generator” in Keras is on Github. It’s not that hard.

Despite the penalty of no-shows, it seems that some presentations were withdrawn. That always bums me out, as it would have been a chance for someone else to go and present.

The Editor’s reception I attended was interesting and honoured some good people. To my surprise the sound system was horrendous. That may be the reason that I missed mentions of the continued effort put forth towards the SEG Wiki, or the changing publishing landscape that shifts towards openness and preprints. I have been IP blocked loading several times loading the SEG abstracts and my abstract is only available via my website as preprints are not allowed on a non-personal website like the ArXiv. A fact I still hope will change soon.

The Weird

Let’s jolly up and talk about the fun and odd things. Lenovo was nice enough to turn the Digital Arena into a playground but see for yourself.

While fewer people visited, the SEG still supported the SEP program. Many people I met around the place were part of this student programme that would teach leadership to distinguished students. The SEG Seismic Soundoff podcast has published several interviews with these students.

Let me know about your highlights in the comments.
What did I miss?
What should I be glad I missed?
What was the oddest thing you saw?

The following two tabs change content below.
... is a geophysicist by heart. He works at the intersection of machine learning and geoscience. He is the founder of The Way of the Geophysicist and a deep learning enthusiast. Writing mostly about computational geoscience and interesting bits and pieces relevant to post-grad life.

Latest posts by Jesper Dramsch (see all)

Posted in Easy Reading, Science and tagged , , , , , .