Computational Methods for Large-scale Inverse Problems: Data-driven VS Physics-driven or Combined?
Tuesday, October 15, 2019 - 11:20am - 12:05pm
Acoustic- and elastic-waveform inversion is an important and widely used method to reconstruct subsurface velocity images. Waveform inversion is a typical non-linear and ill-posed inverse problem. Existing physics-driven computational methods for solving waveform inversion suffer from the cycle skipping and local minima issues, and not to mention solving waveform inversion is computationally expensive. In this work, we developed real-time data-driven techniques to accurately reconstruct subsurface velocities. Our data-driven inversion approaches are end-to-end frameworks that can generate high-quality velocity images directly from the raw seismic waveform data. A series of numerical experiments are conducted on the synthetic seismic reflection data to evaluate the effectiveness and efficiency of our method. We not only compare our methods with existing physics-driven approaches but also choose some deep learning frameworks as our data-driven baselines. The experiment results show that our methods outperform the physics-driven waveform inversion methods and achieve state-of-the-art performance among data-driven baselines.