Information Theory Considerations in Patch-based Training of Deep Neural Networks on Seismic Time-Series

Jesper Sören Dramsch; Mikael Lüthje

doi:10.3997/2214-4609.201803020

Information Theory Considerations in Patch-based Training of Deep Neural Networks on Seismic Time-Series

Jesper Sören Dramsch, Mikael Lüthje

Department of Applied Mathematics and Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

247 Downloads (Orbit)

Abstract

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.

Original language	English
Title of host publication	Proceedings of the First EAGE/PESGB Workshop on Machine Learning (London2018)
Publisher	European Association of Geoscientists and Engineers
Publication date	2018
Pages	46-48
ISBN (Print)	978-1-5108-7668-2
DOIs	https://doi.org/10.3997/2214-4609.201803020
Publication status	Published - 2018
Event	First EAGE/PESGB Workshop Machine Learning - Olympia London, London, United Kingdom Duration: 29 Nov 2018 → 30 Nov 2018

Conference

Conference	First EAGE/PESGB Workshop Machine Learning
Location	Olympia London
Country/Territory	United Kingdom
City	London
Period	29/11/2018 → 30/11/2018

Access to Document

10.3997/2214-4609.201803020

Information Theory Considerations in Patch based Training of Deep Neural Networks on Seismic Time SeriesFinal published version, 332 KB

OpenUrl availability

Full text

Cite this

@inproceedings{8d0b55533543454ab495bb42ae3429a9,

title = "Information Theory Considerations in Patch-based Training of Deep Neural Networks on Seismic Time-Series",

abstract = "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. ",

author = "{S{\"o}ren Dramsch}, Jesper and Mikael L{\"u}thje",

year = "2018",

doi = "10.3997/2214-4609.201803020",

language = "English",

isbn = "978-1-5108-7668-2",

pages = "46--48",

booktitle = "Proceedings of the First EAGE/PESGB Workshop on Machine Learning (London2018)",

publisher = "European Association of Geoscientists and Engineers",

note = "First EAGE/PESGB Workshop Machine Learning ; Conference date: 29-11-2018 Through 30-11-2018",

}

Sören Dramsch, J & Lüthje, M 2018, Information Theory Considerations in Patch-based Training of Deep Neural Networks on Seismic Time-Series. in Proceedings of the First EAGE/PESGB Workshop on Machine Learning (London2018). European Association of Geoscientists and Engineers, pp. 46-48, First EAGE/PESGB Workshop Machine Learning , London, United Kingdom, 29/11/2018. https://doi.org/10.3997/2214-4609.201803020

Information Theory Considerations in Patch-based Training of Deep Neural Networks on Seismic Time-Series. / Sören Dramsch, Jesper; Lüthje, Mikael.
Proceedings of the First EAGE/PESGB Workshop on Machine Learning (London2018). European Association of Geoscientists and Engineers, 2018. p. 46-48.

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

TY - GEN

T1 - Information Theory Considerations in Patch-based Training of Deep Neural Networks on Seismic Time-Series

AU - Sören Dramsch, Jesper

AU - Lüthje, Mikael

PY - 2018

Y1 - 2018

N2 - 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.

AB - 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.

U2 - 10.3997/2214-4609.201803020

DO - 10.3997/2214-4609.201803020

M3 - Article in proceedings

SN - 978-1-5108-7668-2

SP - 46

EP - 48

BT - Proceedings of the First EAGE/PESGB Workshop on Machine Learning (London2018)

PB - European Association of Geoscientists and Engineers

T2 - First EAGE/PESGB Workshop Machine Learning

Y2 - 29 November 2018 through 30 November 2018

ER -

Information Theory Considerations in Patch-based Training of Deep Neural Networks on Seismic Time-Series

Abstract

Conference

Access to Document

OpenUrl availability

Fingerprint

Cite this