Implicit seismic full waveform inversion with deep neural representation

We propose the implicit full waveform inversion (IFWI) algorithm using continuously and implicitly defined deep neural representations. Compared to the conventional full waveform inversion (FWI), which is sensitive to the initial model, IFWI benefits from the increased degrees of freedom and the frequency bias with deep learning optimization, thus allowing to start from a random initialization, which greatly reduces the risk of non-uniqueness and being trapped in local minima. In addition, uncertainty analysis of IFWI can be easily performed by approximating Bayesian inference with various deep learning approaches, such as adding dropout neurons, using deep ensembles, or adopting Beyesian neural networks. Synthetic examples demonstrate that IFWI is able to produce a high-resolution image of subsurface with fine structures, and has strong generalization ability and a certain degree of robustness, which appears to be a well-suited tool for multi-scale joint geophysical inversion.