Reproduce “ML-misfit: Learning a robust misfit function for full-waveform inversion using machine learning” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Bingbing Sun, Tariq Alkhalifah

The authors introduce a learned misfit function for full-waveform inversion (FWI): a neural network scores differences between predicted and recorded seismic traces within the inversion loop. They train the score through tasks with known models, rather than predicting velocities directly. On a modified Marmousi synthetic test with 3–7 Hz data and no frequency continuation, it improves on L2; field performance and general robustness remain unestablished.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.