Industry Insights | Deep Learning at SEG 2020

Machine learning may yet help geophysicists make better decisions, based on more data, and in less time. Andrew Long looks beyond the hype, to identify the progress being made.

Machine learning (ML) dominated the technical program at the 2020 Society of Exploration Geophysicists (SEG). Deep Learning applications were the most popular.

In this edition of Industry Insights, Andrew considers efforts in convolutional neural networks (CNNs) and generative adversarial networks (GANs), and how a combination of these may provide a foundation to evolve ML solutions in the geosciences.

Complementary deep learning efforts may also provide a means to address the lack of very low-frequency signals in the recorded data, resolving one of the key weaknesses in Full Waveform Inversion.