Workshop Description

During the last couple of years, Artificial Intelligence applications have appeared in every corner of geoscience. Many authors have demonstrated that the ML toolbox and deep neural network can be applied successfully to a wide variety of relatively small-scale and well-curated problems. The challenge is now for these ML/AI workflows to mature and be incorporated in a messy production environment. This workshop aims to focus not on proof-of-concept demonstrations or toy problems, but instead on deployment challenges such as incorporation of ML/AI workflows into production toolboxes, continuous training-deployment cycle, implementation challenges relating to deployment, cloud computing, etc.

---

Workshop Programme

09:00	Welcome and Keynote
Session One: Seismic DL Applications in Production
09:30	Deployment of Machine Learning Solutions to Production Seismic Processing R. Hegge (Aramco)
10:00	Experiences in Developing and Deploying Machine Learning for Real-time Processing Joshua Williams (ESG Solutions)
10:30	Q&A
11:00	Coffee Break
Session Two: Engineering Best Practices for ML in Production
11:15	A Principled Approach to Improving Production Readiness  P. Lang (Schlumberger)
11:45	Growing Up: On productizing sub-surface machine learning workflows J. Limbeck (Shell)
12:15	Q&A
12:45	Lunch Break
Session Three: Real Time Analytics and NLP
13:15	Johan Sverdrup Fiber Optics Data Pipeline, Illustrating Why We Need Efficient Workflows for Real-time ML S. Dummong (Equinor)
13:45	Production-scale Processing of EAGE's EarthDoc data to Stimulate New Insights in CO2 and New Energy Management  C. Mamador (Iraya Energies)
14:15	Q&A
14:45	Coffee Break
15:00	Panel Q&A
16:00	End of the Workshop