Model-Based Synthetic Geoelectric Sampling for Magnetotelluric Inversion With Deep Neural Networks

被引:52
作者
Li, Ruiheng [1 ,2 ]
Yu, Nian [1 ,2 ]
Wang, Xuben [3 ]
Liu, Yang [1 ,2 ]
Cai, Zhikun [1 ,2 ]
Wang, Enci [1 ,2 ]
机构
[1] Chongqing Univ, Sch Elect Engn, Chongqing 400044, Peoples R China
[2] Chongqing Univ, State Key Lab Power Transmiss Equipment & Syst Se, Chongqing 400044, Peoples R China
[3] Chengdu Univ Technol, State Key Lab Oil & Gas Reservoir Geol & Exploita, Chengdu 610059, Peoples R China
来源
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING | 2022年 / 60卷
基金
中国国家自然科学基金;
关键词
Training; Conductivity; Artificial neural networks; Predictive models; Data models; Neurons; Mathematical model; Data imbalance; deep neural network (DNN); inversion; magnetotelluric (MT); training sample; 3-D INVERSION;
D O I
10.1109/TGRS.2020.3043661
中图分类号
P3 [地球物理学]; P59 [地球化学];
学科分类号
0708 ; 070902 ;
摘要
Neural networks (NNs) are efficient tools for rapidly obtaining geoelectric models to solve magnetotelluric (MT) inversion problems. Training an NN with strong predictive power requires numerous training samples to prevent underfitting. To reduce the computational burden of generating a large number of training samples, this work analyzes the influence of the sample features and distribution on the training effect for an NN and proposes an efficient method of sample generation. This innovative method consists of three steps: 1) geoelectrically simplifying the features; 2) removing unnecessary features on the basis of realistic geological characteristics; and 3) mapping the samples to a higher-dimensional space. Numerical examples based on simple stratified models show that the number of samples can be reduced to below one-millionth of the original number while improving the predictive effect of the NN. The performance and effectiveness for processing more complex structures are verified by the inversion results obtained for a public data set, COPROD2. We conclude that this advanced method can generate high-quality training samples at a greatly reduced computational cost. The analysis of the sample features and distribution not only advances the state of research on the use of machine learning in geophysical inversion but also is a forward-looking study on the mechanisms of underfitting, tracing the source of these phenomena back to the training samples used.
引用
收藏
页数:14
相关论文
共 45 条
[1]  
Alain Guillaume, 2015, Variance reduction in sgd by distributed importance sampling
[2]   Magnetotelluric inversion of one- and two-dimensional synthetic data based on hybrid genetic algorithms [J].
Batista, Joelson da Conceicao ;
Starteri Sampaio, Edson Emanoel .
ACTA GEOPHYSICA, 2019, 67 (05) :1365-1377
[3]  
Bjerrum E.J., 2017, SMILES ENUMERATION D
[4]  
Borovykh A., 2019, ARXIV190205312
[5]   A systematic study of the class imbalance problem in convolutional neural networks [J].
Buda, Mateusz ;
Maki, Atsuto ;
Mazurowski, Maciej A. .
NEURAL NETWORKS, 2018, 106 :249-259
[6]   Review paper: Instrumentation for marine magnetotelluric and controlled source electromagnetic sounding [J].
Constable, Steven .
GEOPHYSICAL PROSPECTING, 2013, 61 :505-532
[7]   Inverting magnetotelluric responses in a three-dimensional earth using fast forward approximations based on artificial neural networks [J].
Conway, Dennis ;
Alexander, Bradley ;
King, Michael ;
Heinson, Graham ;
Kee, Yang .
COMPUTERS & GEOSCIENCES, 2019, 127 :44-52
[8]   Mean Absolute Percentage Error for regression models [J].
de Myttenaere, Arnaud ;
Golden, Boris ;
Le Grand, Benedicte ;
Rossi, Fabrice .
NEUROCOMPUTING, 2016, 192 :38-48
[9]  
Fawaz H. I., 2018, ARXIV180802455
[10]   FocusNet: Imbalanced Large and Small Organ Segmentation with an End-to-End Deep Neural Network for Head and Neck CT Images [J].
Gao, Yunhe ;
Huang, Rui ;
Chen, Ming ;
Wang, Zhe ;
Deng, Jincheng ;
Chen, Yuanyuan ;
Yang, Yiwei ;
Zhang, Jie ;
Tao, Chanjuan ;
Li, Hongsheng .
MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION - MICCAI 2019, PT III, 2019, 11766 :829-838