Solving inverse wave scattering with deep learning

This paper proposes a neural network approach for solving two classical problems in the two-dimensional inverse wave scattering: far field pattern problem and seismic imaging. The mathematical of scattering is to recover scatterer medium based on boundary measurement scattered from medium, which high-dimensional nonlinear. For under circular experimental setup, perturbative analysis shows that forward map can be approximated by vectorized convolution operator angular direction. Motivated this filtered back-projection, we propose an effective architecture using recently introduced BCR-Net along with standard layers. Analogously imaging problem, similar rectangular domain setup depth-dependent background velocity. Numerical results demonstrate efficiency proposed networks.