Aarhus University Seal / Aarhus Universitets segl

Bayesian Convolutional Neural Networks for Seismic Facies Classification

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avisTidsskriftartikelForskningpeer review

Standard

Bayesian Convolutional Neural Networks for Seismic Facies Classification. / Feng, Runhai; Balling, Niels; Grana, Dario; Dramsch, Jesper Soren; Hansen, Thomas Mejer.

I: IEEE Transactions on Geoscience and Remote Sensing, Bind 59, Nr. 10, 10.2021, s. 8933-8940.

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avisTidsskriftartikelForskningpeer review

Harvard

Feng, R, Balling, N, Grana, D, Dramsch, JS & Hansen, TM 2021, 'Bayesian Convolutional Neural Networks for Seismic Facies Classification', IEEE Transactions on Geoscience and Remote Sensing, bind 59, nr. 10, s. 8933-8940. https://doi.org/10.1109/TGRS.2020.3049012

APA

Feng, R., Balling, N., Grana, D., Dramsch, J. S., & Hansen, T. M. (2021). Bayesian Convolutional Neural Networks for Seismic Facies Classification. IEEE Transactions on Geoscience and Remote Sensing, 59(10), 8933-8940. https://doi.org/10.1109/TGRS.2020.3049012

CBE

Feng R, Balling N, Grana D, Dramsch JS, Hansen TM. 2021. Bayesian Convolutional Neural Networks for Seismic Facies Classification. IEEE Transactions on Geoscience and Remote Sensing. 59(10):8933-8940. https://doi.org/10.1109/TGRS.2020.3049012

MLA

Feng, Runhai o.a.. "Bayesian Convolutional Neural Networks for Seismic Facies Classification". IEEE Transactions on Geoscience and Remote Sensing. 2021, 59(10). 8933-8940. https://doi.org/10.1109/TGRS.2020.3049012

Vancouver

Feng R, Balling N, Grana D, Dramsch JS, Hansen TM. Bayesian Convolutional Neural Networks for Seismic Facies Classification. IEEE Transactions on Geoscience and Remote Sensing. 2021 okt;59(10):8933-8940. https://doi.org/10.1109/TGRS.2020.3049012

Author

Feng, Runhai ; Balling, Niels ; Grana, Dario ; Dramsch, Jesper Soren ; Hansen, Thomas Mejer. / Bayesian Convolutional Neural Networks for Seismic Facies Classification. I: IEEE Transactions on Geoscience and Remote Sensing. 2021 ; Bind 59, Nr. 10. s. 8933-8940.

Bibtex

@article{c305f4248a764a8bb95bf7d7363a65a6,
title = "Bayesian Convolutional Neural Networks for Seismic Facies Classification",
abstract = "The seismic response of geological reservoirs is a function of the elastic properties of porous rocks, which depends on rock types, petrophysical features, and geological environments. Such rock characteristics are generally classified into geological facies. We propose to use the convolutional neural networks in a Bayesian framework to predict facies based on seismic data and quantify the uncertainty in the classification. A variational approach is adopted to approximate the posterior distribution of neural parameters that is mathematically intractable. The network is trained on labeled examples. The mean and the standard deviation of the distribution of neural parameters are randomly drawn from predefined Gaussian functions for the initialization, and are updated by minimizing the negative evidence lower bound. The facies classification is applied to seismic sections not included in the training data set. We draw multiple random samples from the trained variational posterior distribution to simulate an ensemble predictor and classify the most probable seismic facies. We implement the proposed network in the open-source library of TensorFlow Probability, for its convenience and flexibility. The applications show that the internal regions of the seismic sections are generally classified with higher confidence than their boundaries, as measured by the predictive entropy that is calculated based on a multiclass probability across the possible facies. A plain neural network is also applied for comparison, by assigning fixed values to the neural parameters using a classical backpropagation technique. The comparison shows consistent results; however, the proposed approach is able to assess the uncertainty in the predictions.",
keywords = "Bayes methods, Bayesian convolutional neural networks, Deep learning, Monte Carlo methods, Rocks, seismic facies classification, Training, Training data, Uncertainty, uncertainty quantification, variational approach.",
author = "Runhai Feng and Niels Balling and Dario Grana and Dramsch, {Jesper Soren} and Hansen, {Thomas Mejer}",
year = "2021",
month = oct,
doi = "10.1109/TGRS.2020.3049012",
language = "English",
volume = "59",
pages = "8933--8940",
journal = "I E E E Transactions on Geoscience and Remote Sensing",
issn = "0196-2892",
publisher = "Institute of Electrical and Electronics Engineers",
number = "10",

}

RIS

TY - JOUR

T1 - Bayesian Convolutional Neural Networks for Seismic Facies Classification

AU - Feng, Runhai

AU - Balling, Niels

AU - Grana, Dario

AU - Dramsch, Jesper Soren

AU - Hansen, Thomas Mejer

PY - 2021/10

Y1 - 2021/10

N2 - The seismic response of geological reservoirs is a function of the elastic properties of porous rocks, which depends on rock types, petrophysical features, and geological environments. Such rock characteristics are generally classified into geological facies. We propose to use the convolutional neural networks in a Bayesian framework to predict facies based on seismic data and quantify the uncertainty in the classification. A variational approach is adopted to approximate the posterior distribution of neural parameters that is mathematically intractable. The network is trained on labeled examples. The mean and the standard deviation of the distribution of neural parameters are randomly drawn from predefined Gaussian functions for the initialization, and are updated by minimizing the negative evidence lower bound. The facies classification is applied to seismic sections not included in the training data set. We draw multiple random samples from the trained variational posterior distribution to simulate an ensemble predictor and classify the most probable seismic facies. We implement the proposed network in the open-source library of TensorFlow Probability, for its convenience and flexibility. The applications show that the internal regions of the seismic sections are generally classified with higher confidence than their boundaries, as measured by the predictive entropy that is calculated based on a multiclass probability across the possible facies. A plain neural network is also applied for comparison, by assigning fixed values to the neural parameters using a classical backpropagation technique. The comparison shows consistent results; however, the proposed approach is able to assess the uncertainty in the predictions.

AB - The seismic response of geological reservoirs is a function of the elastic properties of porous rocks, which depends on rock types, petrophysical features, and geological environments. Such rock characteristics are generally classified into geological facies. We propose to use the convolutional neural networks in a Bayesian framework to predict facies based on seismic data and quantify the uncertainty in the classification. A variational approach is adopted to approximate the posterior distribution of neural parameters that is mathematically intractable. The network is trained on labeled examples. The mean and the standard deviation of the distribution of neural parameters are randomly drawn from predefined Gaussian functions for the initialization, and are updated by minimizing the negative evidence lower bound. The facies classification is applied to seismic sections not included in the training data set. We draw multiple random samples from the trained variational posterior distribution to simulate an ensemble predictor and classify the most probable seismic facies. We implement the proposed network in the open-source library of TensorFlow Probability, for its convenience and flexibility. The applications show that the internal regions of the seismic sections are generally classified with higher confidence than their boundaries, as measured by the predictive entropy that is calculated based on a multiclass probability across the possible facies. A plain neural network is also applied for comparison, by assigning fixed values to the neural parameters using a classical backpropagation technique. The comparison shows consistent results; however, the proposed approach is able to assess the uncertainty in the predictions.

KW - Bayes methods

KW - Bayesian convolutional neural networks

KW - Deep learning

KW - Monte Carlo methods

KW - Rocks

KW - seismic facies classification

KW - Training

KW - Training data

KW - Uncertainty

KW - uncertainty quantification

KW - variational approach.

UR - http://www.scopus.com/inward/record.url?scp=85100493547&partnerID=8YFLogxK

U2 - 10.1109/TGRS.2020.3049012

DO - 10.1109/TGRS.2020.3049012

M3 - Journal article

AN - SCOPUS:85100493547

VL - 59

SP - 8933

EP - 8940

JO - I E E E Transactions on Geoscience and Remote Sensing

JF - I E E E Transactions on Geoscience and Remote Sensing

SN - 0196-2892

IS - 10

ER -