An Interview with Energy Solutions VP Mason Dykstra
AI/Machine Learning Challenges Facing Energy Industry Operators
In this interview conducted by Upstream Content Studio, Mason provides an array of insights and examples on the challenges for operators in taking advantage of the rapidly advancing capabilities of AI/machine learning. Among the topics:
- The commoditization of machine learning models
- Industry and domain specific data sets analogous to ImageNet
- The importance of making the AI/ML technology invisible to the expert
- Advice for operators on partnering to realize the potential of AI/ML
- The concept of ‘applied digital innovation’ in finding the way forward
In this 35 minute interview, Energy Solutions VP Mason Dykstra provides insights covering a range of topics on the challenges facing operators for applying AI/machine learning in their business, from ensuring early and ongoing value to choosing the right projects and partners.
Machine Learning Interprets Like Experts
- Interpret with orders of magnitude greater efficiency.
- Integrate, visualize and develop new AI/machine learning models.
- Gain unique understanding enabled by pattern recognition-based AI.
- Import results into industry-standard software packages, aligned to OSDU.
4m12s | Learn about SubsurfaceAI Seismic, a cloud native application with a web front end, employing pattern recognition-based machine learning models, aligned with the OSDU Data Platform.
Pattern Detection Based AI Models
In this 1-minute video, geoscientist and VP Energy Solutions Mason Dykstra sets out how a new generation of machine learning models based on pattern recognition enable seismic experts to interpret using their human intuition.
Introducing Pattern Recognition AI
In this 5-minute video, Ben Lasscock, Energy Solutions Group Technical Lead, provides a deep dive into the SubsurfaceAI custom deep learning application for seismic, from labeling just a few lines and QCing predictions, to a final interpretation of the data set.
AI/Machine Learning in Geoscience Updates
Empowering Geoscience Experts
Sign up for periodic emails from the Enthought Energy Team for insights on gaining a deep understanding from your data using the most advanced AI/machine learning techniques.
Scientific Software Technologies at Work
See case studies where client domain experts collaborated with Enthought scientists to develop customized software applications taking on their most challenging problems.
You know seismic. We know Python.
Today, there are two major barriers to achieving the potential of deep learning in seismic interpretation; being able to quickly and consistently label seismic data, and being able to access (and use) a deep learning toolkit that enables quick experimentation with models, while iterating and managing all the data associated with your model and results.
Geophysicists need to be highly efficient at producing labeled data to train (and later, iteratively QC) deep learning models to deliver high quality interpretations.
SubsurfaceAI Seismic overcomes both barriers. With as few as three interpreted seismic lines, the pattern recognition-based deep learning models enable you to create robust predictions across the entire seismic volume.
In this 4-minute video, Ben Lasscock, Energy Solutions Group Technical Lead at Enthought, discusses one of the deep learning elements of SubsurfaceAI Seismic, automating the interpretation of sequence stratigraphy. Connect with Ben on LinkedIn
Explore a New Generation of Digitally Enabled Workflows
Talk to us about using all your subsurface data, removing drudgery from the work of experts, and using the power of today's AI/Machine Learning techniques to get actionable results.