Transfer Learning Enhanced Full Waveform Inversion
This work addresses efficiency issues in non-destructive testing for applications like material inspection, but it appears incremental as it builds on existing FWI methods with neural network enhancements.
The paper tackles the problem of improving efficiency in Full Waveform Inversion (FWI) for non-destructive testing by using pretrained neural networks to provide a good starting point, which reduces the number of iterations in the inversion process.
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.