Learning approaches have revolutionized machine vision, speech recognition, and a host of other domains. However, the development of machine learning algorithms for geophysical problems has yet to be truly explored while, as compared with standard approaches for (supervised) learning, various adjustments have to be made. For example, often "ground truth" data are not available. Nonetheless, machine learning integrating physics is leading to a new era of geophysics research. Here, we show that deep neural network architectures arise naturally in different seismic inverse problems on the one hand, and that learning the resulting networks poses promising new approaches to studying Earth’s subsurface on the other hand.

Presentation Date: Monday, October 15, 2018

Start Time: 1:50:00 PM

Location: 204B (Anaheim Convention Center)

Presentation Type: Oral

Keywords:
generative model, deep generative model, deep learning, hoop, neural network, machine learning, deep neural network architecture, seg international exposition, imaging, generative adversarial network
Subjects:
Information Management and Systems, Neural networks
This content is only available via PDF.
2018. Society of Exploration Geophysicists
You can access this article if you purchase or spend a download.