Abstract:
Recent developments in application of deep learning models to acoustic Full Waveform Inversion (FWI) are marked by the use of diffusion models as prior distributions for Bayesian-like inference procedures. The advantage of these methods is the ability to generate high-resolution samples, which are otherwise unattainable with classical inversion methods or other deep learning-based solutions. However, the iterative and stochastic nature of sampling from diffusion models along with heuristic nature of output control remain limiting factors for their applicability For instance, an optimal way to include the approximate velocity model into diffusion-based inversion scheme remains unclear, even though it is considered an essential part of FWI pipeline. We address the issue by employing a Schrodinger Bridge that interpolates between the distributions of ground truth and smoothed velocity models. Thus, the inference process that starts from an approximate velocity model is guaranteed to arrive at a sample from the distribution of reference velocity models in a finite time. To facilitate the learning of nonlinear drifts that transfer samples between distributions and to enable controlled inference given the seismic data, we extend the concept of Image-to-Image Schrodinger Bridge (I2SB) to conditional sampling, resulting in a conditional Image-to-Image Schrodinger Bridge (cI2SB) framework for acoustic inversion. To validate our method, we assess its effectiveness in reconstructing the reference velocity model from its smoothed approximation, coupled with the observed seismic signal of fixed shape. Our experiments demonstrate that the proposed solution outperforms our reimplementation of conditional diffusion model suggested in earlier works, while requiring only a few neural function evaluations (NFEs) to achieve sample fidelity superior to that attained with supervised learning-based approach. The supplementary code implementing the algorithms described in this paper can be found in the repository.
