This is a Virtual Event. The meeting will begin at 5:00 PM

NOTE: You Must Be Logged In to Register.

Speakers: Dr. Heather Bedle and Dr. David Lubo-Robles, University of Oklahoma

Due to the growing integration of artificial intelligence (AI) techniques in geophysical workflows such as structural and stratigraphic interpretation using seismic reflection data, quantitative seismic interpretation, and seismic processing, questions regarding the optimal input features for machine learning (ML) architectures, assess the uncertainty in the model outputs, and correlate the results with the geology persist. This talk explores the transformative potential of an explainable AI technique called Shapley additive explanations (SHAP) in demystifying the "black box" concept of ML models. Using a random forest architecture trained for seismic facies analysis, we study SHAP's ability to analyze input-to-output relationships to understand how the ML architecture uses a suite of input seismic attributes to perform the classification. Moreover, we extend SHAP's utility beyond traditional seismic analysis by investigating public perceptions of induced seismicity. This exploration of explainable AI to enhance the interpretability of ML architectures for different applications offers deeper insights into effectively integrating AI techniques in geoscience.

Speaker Biographies:
Dr. Heather Bedle, University of Oklahoma
Dr. Heather Bedle is an assistant professor in the School of Geosciences at the University of Oklahoma.  As PI of the Attribute Assisted Seismic Processing & Imaging (AASPI) Consortium, she is passionate about mentoring aspiring scientists and exploring the complex intersections of earth, energy, and environment. Holding a Ph.D. from Northwestern University, she leverages her industry and academic experiences to lead a prolific research group recognized for their interdisciplinary approach, seamlessly combining geoscience and data analytics to tackle pressing scientific questions.

Dr. David Lubo-Robles, University of Oklahoma
Dr. David Lubo-Robles is a Postdoctoral Research Associate in the School of Geosciences at the University of Oklahoma. David received a B.S. in geophysical engineering from Simon Bolivar University, Venezuela, and an M.S. and Ph.D. in geophysics from the University of Oklahoma. Working with the Attribute Assisted Seismic Processing & Imaging (AASPI) Consortium, his research interests include the development and application of innovative tools using machine learning, quantitative interpretation, and seismic attribute analysis to delineate geologic features amenable for energy and climate solutions.

**Access information will be sent to all registrants after registration closes.

Price List:

** Please be aware that ALL registrations must go through the 'Checkout', even if your total is zero. Registration is not confirmed until you complete the checkout process.