Initial Experiments on Improving Seismic Data Inversion with Deep Learning

摘要

The main objective of the work is to explore if deep learning can facilitate the seismic data inversion problem. The paper describes the challenges of seismic Full Waveform Inversion (FWI) process, and presents our method of creating a deep learning model into the FWI workflow. The paper presents the initial findings of the research in enhancing the FWI iterative optimization process. These initial findings demonstrate that deep learning is able to help physics-based solution by improving its numerical optimization.