A deep learning enhanced framework for multi-physics joint inversion

What is it about?

This paper describes a deep learning enhanced (DLE) framework designed to combine the DNN and the traditional separate inversion workflows together and working in the model domain to improve the multi-physics joint inversion results iteratively.

Featured Image

Why is it important?

Our framework has great flexibility to deal with different sensing configurations and nonconforming discretization, and excellent generalization abilities when tested on datasets using divergent geological structures.