Transfer Learning Enhanced Full Waveform Inversion

TL;DR

This paper introduces a transfer learning approach to improve the efficiency of Full Waveform Inversion in non-destructive testing by using pretrained neural networks to initialize the inversion process, reducing computational effort.

Contribution

It presents a novel method combining neural networks with adjoint optimization in FWI, leveraging transfer learning for faster convergence in material property reconstruction.

Findings

Pretrained neural networks significantly reduce inversion iterations.

The method improves efficiency in non-destructive testing applications.

Transfer learning enhances FWI convergence speed.

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.