2000 character limit reached
WISER: multimodal variational inference for full-waveform inversion without dimensionality reduction
Published 3 May 2024 in physics.geo-ph and cs.CE | (2405.10327v2)
Abstract: We present a semi-amortized variational inference framework designed for computationally feasible uncertainty quantification in 2D full-waveform inversion to explore the multimodal posterior distribution without dimensionality reduction. The framework is called WISER, short for full-Waveform variational Inference via Subsurface Extensions with Refinements. WISER leverages the power of generative artificial intelligence to perform approximate amortized inference that is low-cost albeit showing an amortization gap. This gap is closed through non-amortized refinements that make frugal use of acoustic wave physics. Case studies illustrate that WISER is capable of full-resolution, computationally feasible, and reliable uncertainty estimates of velocity models and imaged reflectivities.
- J. Virieux and S. Operto. An overview of full-waveform inversion in exploration geophysics. GEOPHYSICS, 74(6):WCC1–WCC26, 11 2009. doi: 10.1190/1.3238367. URL http://dx.doi.org/10.1190/1.3238367.
- The frontier of simulation-based inference. Proceedings of the National Academy of Sciences, 117(48):30055–30062, 2020.
- Uncertainty quantification for inverse problems with weak partial-differential-equation constraints. Geophysics, 83(6):R629–R647, 2018.
- Null-space shuttles for targeted uncertainty analysis in full-waveform inversion. Geophysics, 86(1):R63–R76, 2021.
- Physics reliable frugal uncertainty analysis for full waveform inversion. arXiv preprint arXiv:2305.07921, 2023.
- Local uncertainty quantification for 3-d time-domain full-waveform inversion with ensemble kalman filters: application to a north sea obc data set. Geophysical Journal International, 237(3):1353–1383, 2024.
- Markov chain monte carlo convergence diagnostics: a comparative review. Journal of the American statistical Association, 91(434):883–904, 1996.
- Deep generator priors for bayesian seismic inversion. In Advances in geophysics, volume 61, pages 179–216. Elsevier, 2020.
- Uncertainty quantification of geologic model parameters in 3d gravity inversion by hessian-informed markov chain monte carlo. Geophysics, 88(1):G1–G18, 2023.
- Quantifying uncertainty of salt body shapes recovered from gravity data using trans-dimensional markov chain monte carlo sampling. Geophysical Journal International, 232(3):1957–1978, 2023.
- 3d monte carlo geometry inversion using gravity data. Geophysics, 89(3):1–62, 2024a.
- Reconstruction of multiple target bodies using trans-dimensional bayesian inversion with different constraints. IEEE Transactions on Geoscience and Remote Sensing, 2024b.
- Hamiltonian monte carlo based elastic full-waveform inversion of wide-angle seismic data. Geophysical Journal International, 237(3):1384–1399, 2024.
- An introduction to variational inference in geophysical inverse problems. In Advances in geophysics, volume 62, pages 73–140. Elsevier, 2021.
- Conditional score-based diffusion models for bayesian inference in infinite dimensions. Advances in Neural Information Processing Systems, 36, 2024.
- Learning likelihoods with conditional normalizing flows. arXiv preprint arXiv:1912.00042, 2019.
- Reliable amortized variational inference with physics-based latent distribution correction. GEOPHYSICS, 88(3):R297–R322, 04 2023. doi: 10.1190/geo2022-0472.1. URL http://dx.doi.org/10.1190/geo2022-0472.1.
- Amortized normalizing flows for transcranial ultrasound with uncertainty quantification. In Medical Imaging with Deep Learning, 2023a. URL https://openreview.net/forum?id=LoJG-lUIlk.
- Wise: full-waveform variational inference via subsurface extensions. Geophysics, 89(4):1–31, 2024.
- Iterative amortized inference. In International Conference on Machine Learning, pages 3403–3412. PMLR, 2018.
- Bayesian seismic tomography using normalizing flows. Geophysical Journal International, 228(1):213–239, 2022.
- 3-d bayesian variational full waveform inversion. Geophysical Journal International, 234(1):546–561, 2023.
- Bayesian variational time-lapse full waveform inversion. Geophysical Journal International, 237(3):1624–1638, 2024.
- Hint: Hierarchical invertible neural transport for density estimation and bayesian inference. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, pages 8191–8199, 2021.
- Building complex synthetic models to evaluate acquisition geometries and velocity inversion technologies. In 74th EAGE Conference and Exhibition incorporating EUROPEC 2012, pages cp–293. European Association of Geoscientists & Engineers, 2012.
- Accelerating extended least-squares migration with weighted conjugate gradient iteration. GEOPHYSICS, 81(4):S165–S179, 07 2016. doi: 10.1190/geo2015-0499.1. URL http://dx.doi.org/10.1190/geo2015-0499.1.
- Fons ten Kroode. An omnidirectional seismic image extension. Inverse Problems, 39(3):035003, 2023.
- Bayesflow: Learning complex stochastic models with invertible neural networks. IEEE transactions on neural networks and learning systems, 33(4):1452–1466, 2020.
- Adjoint operators enable fast and amortized machine learning based bayesian uncertainty quantification. Medical Imaging 2023: Image Processing, 04 2023b. doi: 10.1117/12.2651691. URL http://dx.doi.org/10.1117/12.2651691.
- Weak deep priors for seismic imaging. In SEG Technical Program Expanded Abstracts 2020, pages 2998–3002. Society of Exploration Geophysicists, 2020.
- Total variation regularization strategies in full-waveform inversion. SIAM Journal on Imaging Sciences, 11(1):376–406, 2018.
- Projection methods and applications for seismic nonlinear inverse problems with multiple constraints. Geophysics, 84(2):R251–R269, 2019.
- Multiscale seismic waveform inversion. Geophysics, 60(5):1457–1473, 1995.
- Simply denoise: Wavefield reconstruction via jittered undersampling. Geophysics, 73(3):V19–V28, 2008.
- Frugal full-waveform inversion: From theory to a practical algorithm. The Leading Edge, 32(9):1082–1092, 2013.
- Invertiblenetworks.jl: A julia package for scalable normalizing flows. arXiv preprint arXiv:2312.13480, 2023c.
- Seismic foundation model (sfm): a new generation deep learning model in geophysics. arXiv preprint arXiv:2309.02791, 2023.
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.
Top Community Prompts
Collections
Sign up for free to add this paper to one or more collections.