Show simple item record

dc.contributor.authorSun, Bingbing
dc.contributor.authorAlkhalifah, Tariq Ali
dc.date.accessioned2021-02-08T06:23:50Z
dc.date.available2021-02-08T06:23:50Z
dc.date.issued2020-10-21
dc.date.submitted2019-09-27
dc.identifier.citationSun, B., & Alkhalifah, T. (2020). ML-descent: An optimization algorithm for full-waveform inversion using machine learning. GEOPHYSICS, 85(6), R477–R492. doi:10.1190/geo2019-0641.1
dc.identifier.issn0016-8033
dc.identifier.issn1942-2156
dc.identifier.doi10.1190/geo2019-0641.1
dc.identifier.urihttp://hdl.handle.net/10754/667268
dc.description.abstractFull-waveform inversion (FWI) is a nonlinear optimization problem, and a typical optimization algorithm such as the nonlinear conjugate gradient or limited-memory Broyden-Fletcher-Goldfarb-Shanno (LBFGS) would iteratively update the model mainly along the gradient-descent direction of the misfit function or a slight modification of it. Based on the concept of meta-learning, rather than using a hand-designed optimization algorithm, we have trained the machine (represented by a neural network) to learn an optimization algorithm, entitled the “ML-descent,” and apply it in FWI. Using a recurrent neural network (RNN), we use the gradient of the misfit function as the input, and the hidden states in the RNN incorporate the history information of the gradient similar to an LBFGS algorithm. However, unlike the fixed form of the LBFGS algorithm, the machine-learning (ML) version evolves in response to the gradient. The loss function for training is formulated as a weighted summation of the L2 norm of the data residuals in the original inverse problem. As with any well-defined nonlinear inverse problem, the optimization can be locally approximated by a linear convex problem; thus, to accelerate the training, we train the neural network by minimizing randomly generated quadratic functions instead of performing time-consuming FWIs. To further improve the accuracy and robustness, we use a variational autoencoder that projects and represents the model in latent space. We use the Marmousi and the overthrust examples to demonstrate that the ML-descent method shows faster convergence and outperforms conventional optimization algorithms. The energy in the deeper part of the models can be recovered by the ML-descent even when the pseudoinverse of the Hessian is not incorporated in the FWI update.
dc.description.sponsorshipWe thank the editor-in-chief J. Shragge for improving the manuscript. We appreciate the comments and suggestions by the assistant editor A. Guitton and the associate editor S. A. L. de Ridder in the reviewing process. We thank the members of SWAG at KAUST for the useful discussions.
dc.publisherSociety of Exploration Geophysicists
dc.relation.urlhttp://mr.crossref.org/iPage?doi=10.1190%2Fgeo2019-0641.1
dc.rightsArchived with thanks to GEOPHYSICS
dc.titleML-descent: An optimization algorithm for full-waveform inversion using machine learning
dc.typeArticle
dc.contributor.departmentPhysical Science and Engineering (PSE) Division
dc.contributor.departmentEarth Science and Engineering Program
dc.identifier.journalGEOPHYSICS
dc.eprint.versionPost-print
dc.identifier.volume85
dc.identifier.issue6
dc.identifier.pagesR477-R492
kaust.personSun, Bingbing
kaust.personAlkhalifah, Tariq Ali
dc.date.accepted2020-06-21
refterms.dateFOA2021-02-08T10:55:28Z
kaust.acknowledged.supportUnitmembers of SWAG


Files in this item

Thumbnail
Name:
ML_descent_compressed-2.pdf
Size:
1.434Mb
Format:
PDF
Description:
Accepted manuscript

This item appears in the following Collection(s)

Show simple item record