Current browse context:
eess
Change to browse by:
References & Citations
Physics > Computational Physics
Title: Accelerating Least Squares Imaging Using Deep Learning Techniques
(Submitted on 14 Nov 2019 (v1), last revised 10 Dec 2019 (this version, v2))
Abstract: Wave equation techniques have been an integral part of geophysical imaging workflows to investigate the Earth's subsurface. Least-squares reverse time migration (LSRTM) is a linearized inversion problem that iteratively minimizes a misfit functional as a function of the model perturbation. The success of the inversion largely depends on our ability to handle large systems of equations given the massive computation costs. The size of the system almost exponentially increases with the demand for higher resolution images in complicated subsurface media. We propose an unsupervised deep learning approach that leverages the existing physics-based models and machine learning optimizers to achieve more accurate and cheaper solutions. We compare different optimizers and demonstrate their efficacy in mitigating imaging artifacts. Further, minimizing the Huber loss with mini-batch gradients and Adam optimizer is not only less memory-intensive but is also more robust. Our empirical results on synthetic, densely sampled datasets suggest faster convergence to an accurate LSRTM result than a traditional approach.
Submission history
From: Janaki Vamaraju [view email][v1] Thu, 14 Nov 2019 10:41:49 GMT (5787kb,D)
[v2] Tue, 10 Dec 2019 01:50:10 GMT (5787kb,D)
Link back to: arXiv, form interface, contact.