Deep learning parameterization for geophysical inverse problems

PreviousNext No AccessSEG 2019 Workshop: Mathematical Geophysics: Traditional vs Learning, Beijing, China, 5–7 November 2019Deep learning parameterization for geophysical inverse problemsAuthors: Yunzhi ShiXinming WuSergey FomelYunzhi ShiThe University of Texas at AustinSearch for more papers by this author, Xinming WuThe University of Texas at AustinSearch for more papers by this author, and Sergey FomelThe University of Texas at AustinSearch for more papers by this authorhttps://doi.org/10.1190/iwmg2019_09.1 SectionsAboutPDF/ePub ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InRedditEmail Abstract We present a novel regularization framework that parameterizes the model with a random-initialized convolutional neural network (CNN). The deep network is optimized iteratively and can constrain the space of solutions to admissible models. The method yields coherent and natural-looking models, however, does not require any shaping operator designed with expertise knowledge. A significant improvement is observed using the proposed regularization in comparison to the traditional methods in inverse problems such as noise attenuation and seismic interpolation. Keywords: attenuation, interpolation, inversion, neural networks, reconstructionPermalink: https://doi.org/10.1190/iwmg2019_09.1FiguresReferencesRelatedDetailsCited bySeismic reflectivity inversion via a regularized deep image priorHongling Chen, Mauricio D. Sacchi, and Jinghuai Gao15 August 2022Unsupervised deep learning with higher-order total-variation regularization for multidimensional seismic data reconstructionThomas André Larsen Greiner, Jan Erik Lie, Odd Kolbjørnsen, Andreas Kjelsrud Evensen, Espen Harris Nilsen, Hao Zhao, Vasily Demyanov, and Leiv-J. Gelius31 December 2021 | GEOPHYSICS, Vol. 87, No. 2Uncertainty quantification in imaging and automatic horizon tracking — A Bayesian deep-prior based approachAli Siahkoohi, Gabrio Rizzuti, and Felix J. Herrmann30 September 2020Weak deep priors for seismic imagingAli Siahkoohi, Gabrio Rizzuti, and Felix J. Herrmann30 September 2020Seismic data interpolation using a POCS-guided deep image priorMin Jun Park, Joseph Jennings, Bob Clapp, and Biondo Biondi30 September 2020 SEG 2019 Workshop: Mathematical Geophysics: Traditional vs Learning, Beijing, China, 5–7 November 2019ISSN (online):2159-6832Copyright: 2020 Pages: 138 publication data© 2020 Published in electronic format with permission by the Society of Exploration GeophysicistsPublisher:Society of Exploration Geophysicists HistoryPublished Online: 17 Jan 2020 CITATION INFORMATION Yunzhi Shi, Xinming Wu, and Sergey Fomel, (2020), "Deep learning parameterization for geophysical inverse problems," SEG Global Meeting Abstracts : 36-40. https://doi.org/10.1190/iwmg2019_09.1 Plain-Language Summary Keywordsattenuationinterpolationinversionneural networksreconstructionPDF DownloadLoading ...