Transfer Learning Enhanced Full Waveform Inversion
Machine Learning
2023-12-04 v2 Computational Physics
Geophysics
Abstract
We propose a way to favorably employ neural networks in the field of non-destructive testing using Full Waveform Inversion (FWI). The presented methodology discretizes the unknown material distribution in the domain with a neural network within an adjoint optimization. To further increase efficiency of the FWI, pretrained neural networks are used to provide a good starting point for the inversion. This reduces the number of iterations in the Full Waveform Inversion for specific, yet generalizable settings.
Cite
@article{arxiv.2302.11259,
title = {Transfer Learning Enhanced Full Waveform Inversion},
author = {Stefan Kollmannsberger and Divya Singh and Leon Herrmann},
journal= {arXiv preprint arXiv:2302.11259},
year = {2023}
}
Comments
7 pages, 5 figures