Full text loading...
-
Information Theory Considerations In Patch-Based Training Of Deep Neural Networks On Seismic Time-Series
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, First EAGE/PESGB Workshop Machine Learning, Nov 2018, Volume 2018, p.1 - 3
Abstract
Summary
Recent advances in machine learning relies on convolutional deep neural networks. These are often trained on cropped image patches. Pertaining to non-stationary seismic signals this may introduce low frequency noise and non-generalizability.
© EAGE Publications BV