Practical deep learning inversion using neural architecture search and a flexible training dataset generator

T. Shibayama, N. Mizuno, H. Kusano, A. Kinoshita, M. Minegishi, R. Sakamoto, K. Hasegawa, F. Kachi
{"title":"Practical deep learning inversion using neural architecture search and a flexible training dataset generator","authors":"T. Shibayama, N. Mizuno, H. Kusano, A. Kinoshita, M. Minegishi, R. Sakamoto, K. Hasegawa, F. Kachi","doi":"10.3997/2214-4609.202112777","DOIUrl":null,"url":null,"abstract":"Summary Deep learning has the potential to estimate velocity models directly from shot gathers, which would reduce the turn-around time of seismic inversion. Our study addresses two challenges in implementing deep learning techniques for seismic inversion: the practical generation of a large amount of training data and the search for the best neural network architecture. First, we propose a flexible system which parametrically generates velocity models to create a large-scale, complex and fully synthetic training dataset, without using a target subsurface model. Using this system, we created 300,000 synthetic velocity models for our experiments. Second, we employ neural architecture search techniques to design a suitable neural network using Optuna, an automatic hyperparameter optimisation framework. We incorporated the residual network into an encoder–decoder model and optimised its architecture. Thus, we obtained an optimal neural network model consisting of more than 100 hidden layers. We evaluated our model using the Marmousi2 model and the 1994 Amoco statics test dataset. The model demonstrated comprehensible estimations of the benchmark velocity models.","PeriodicalId":143998,"journal":{"name":"82nd EAGE Annual Conference & Exhibition","volume":"22 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-10-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"82nd EAGE Annual Conference & Exhibition","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.3997/2214-4609.202112777","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

Abstract

Summary Deep learning has the potential to estimate velocity models directly from shot gathers, which would reduce the turn-around time of seismic inversion. Our study addresses two challenges in implementing deep learning techniques for seismic inversion: the practical generation of a large amount of training data and the search for the best neural network architecture. First, we propose a flexible system which parametrically generates velocity models to create a large-scale, complex and fully synthetic training dataset, without using a target subsurface model. Using this system, we created 300,000 synthetic velocity models for our experiments. Second, we employ neural architecture search techniques to design a suitable neural network using Optuna, an automatic hyperparameter optimisation framework. We incorporated the residual network into an encoder–decoder model and optimised its architecture. Thus, we obtained an optimal neural network model consisting of more than 100 hidden layers. We evaluated our model using the Marmousi2 model and the 1994 Amoco statics test dataset. The model demonstrated comprehensible estimations of the benchmark velocity models.
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
使用神经结构搜索和灵活的训练数据集生成器的实用深度学习反演
深度学习有可能直接从射击集估计速度模型,这将减少地震反演的周转时间。我们的研究解决了在地震反演中实施深度学习技术的两个挑战:大量训练数据的实际生成和寻找最佳神经网络架构。首先,我们提出了一个灵活的系统,该系统可以参数化地生成速度模型,以创建大规模、复杂和完全合成的训练数据集,而无需使用目标地下模型。使用这个系统,我们为我们的实验创建了30万个合成速度模型。其次,我们采用神经结构搜索技术,使用Optuna(一个自动超参数优化框架)设计一个合适的神经网络。我们将残差网络整合到编码器-解码器模型中,并对其结构进行了优化。因此,我们得到了一个由100多个隐藏层组成的最优神经网络模型。我们使用Marmousi2模型和1994年Amoco静态测试数据集来评估我们的模型。该模型证明了基准速度模型的可理解估计。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 去求助
来源期刊
自引率
0.00%
发文量
0
期刊最新文献
GOMCRUST - The crustal-scale extension of the 2004 BP velocity model for long-offset OBN acquisition setting Prestack data attenuation compensation based on inversion Complex Near-surface Velocity Modeling via U-net Integrated high-resolution model building: a case study from the Sultanate of Oman Inverse Hessian estimation in least-squares migration using chains of operators
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1