Podcast Tag: Machine learning

Episode 210: Unveiling Seismic Secrets – Inside Machine Learning’s Black Box

“It’s ​not ​like ​machine ​learning ​will ​solve ​all ​the ​problems. ​It’s ​not ​a ​magical ​tool.”

David Lubo-Robles highlights his award-winning paper that utilized novel machine learning methods to enhance interpretability in seismic volume data from the Gulf of Mexico. Listeners will gain insight into the critical role of input quality in machine learning outcomes, the importance of balancing datasets, and the necessity of geoscientific validation. The episode also addresses common misconceptions about machine learning in geophysics, emphasizing the need for critical thinking and geological knowledge to apply these advanced techniques.

Episode 209: Thinking like an algorithm – utilizing machine learning in seismic data

“The driving objective of AASPI is to try and reveal and see more patterns in the seismic data than we can see just looking at the seismic amplitude data.”

Heather Bedle, Principal Investigator at Attribute Assisted Seismic Processing and Interpretation (AASPI) at the University of Oklahoma, joins Seismic Soundoff. In this episode, you will discover how AASPI reveals hidden patterns in seismic data, pushes the boundaries of geologic interpretation, and reshapes our understanding of the Earth using cutting-edge research and technology.

Episode 208: Pioneering Seismic Imaging for Energy and Sustainability

Biondo Biondi, the Director of the Stanford Earth imaging Project (SEP), discusses SEP’s 50-year history and future outlook. Biondo reflects on SEP’s founding during the 1970s oil crisis and today as it tackles modern energy challenges. Biondo discusses how improving seismic imaging can support the future of carbon capture and geothermal energy and help build resilient cities. He also shares why he believes so many SEP alums have been guests on this podcast!

Episode 196: The sound of seismic

Paolo Dell’Aversana highlights his article in The Leading Edge, discussing a dual-sensory approach to understanding seismic. Based on concepts well-established in cognitive sciences, Paolo introduces the idea of expanded imaging in geophysics, using a dual-sensory (audiovisual) perception of a data set. In this episode, Paolo explains the basic principles of multimodal seismic data analysis using augmented imaging theory. He shares the advantages and limitations of converting seismic data into an auditory format and outlines how geophysicists can start with this approach today. This episode unlocks secret information hiding in your seismic data waiting to be discovered.

Episode 194: Improving integration in machine learning workflows

Felix J. Herrmann discusses his open-access article, “Learned multiphysics inversion with differentiable programming and machine learning.” He shares why the future of the oil and gas industry depends on the democratization of technology design. He provides insights into why modernizing wave-equation inversion frameworks is important to geophysics and shares the implications for the results of his study. This episode provides a glimpse into the future capabilities of machine learning to help provide the path for the next great discoveries in geophysics.

Episode 189: How to apply machine learning to real-world problems

Mathematician Herman Jaramillo discusses his new book, Machine Learning for Science and Engineering Volume One: Fundamentals. As the size and complexity of data soars exponentially, ML has gained prominence in applications in geoscience and related fields. ML-powered technology increasingly rivals or surpasses human performance and fuels a large range of leading-edge research. This conversation explores the hottest topics facing students, scientists, and engineers and provides a solid foundation to understand how to utilize this cutting-edge science in your work.

Episode 166: Integrating digital transformation into your business

Steve Darnell discusses how digital transformation improves business processes in-depth. He emphasizes the importance of cybersecurity, how to start the digitalization process, and highlights the common obstacles companies face when embracing digital transformation. He also comments on the common misperceptions and the hidden benefits of embracing digital advancements. This conversation on digital transformation connects to all parts of the oil and gas workflow and showcases the value proposition for companies. 

Episode 161: The Benefits of ML & AI Hinge on a Common Denominator

Chris Hanton, Director of Digital Transformation Solutions at Ikon Science, discusses the latest insights using machine learning and artificial intelligence. Chris highlights one frequently overlooked variable that can undermine the best technology. He also shares how technologists can ensure data is trustworthy and valuable and presents the use case for investing in machine learning and AI outside of increased efficiency. This is a deep dive into the role of quality data in cutting-edge work and the best ways to harness the benefits of machine learning and artificial intelligence.

Episode 159: Optimizing the benefits of machine learning for scientific problems

Souvik Mukherjee discusses his article in July's The Leading Edge about high-resolution imaging of subsurface infrastructure using AI. In this conversation, Souvik presents field study results from Texas and California that show the potential for imaging pipelines and other subsurface infrastructure using AI-based methods on high-resolution aboveground magnetic data. He also highlights the similarities and differences between conventional least-squares inversion and machine learning-based inversion and how he achieved a 100-fold increase in efficiency. Whether exploring AI, machine learning, algorithms, or the latest geophysical technology, this conversation has something for everyone.

Episode 155: Removing the starting model for FWI

Arnab Dhara discusses his paper, "Physics-guided deep autoencoder to overcome the need for a starting model in full-waveform inversion," in the June issue of The Leading Edge. Arnab proposes employing deep learning as a regularization in full-waveform inversion. He explains why physics-based solutions with machine learning are challenging to develop, how he made it possible to train the network without known answers, and why he tested his approach with the Marmousi and SEAM models. Arnab also shares why this research took over 20 years to build on the initial idea and how he used full-waveform inversion without a starting model. This is a cutting-edge conversation that may represent the future of FWI.

ADVERTISEMENT