Abstract:We consider the inverse acoustic obstacle problem for sound-soft star-shaped obstacles in two dimensions wherein the boundary of the obstacle is determined from measurements of the scattered field at a collection of receivers outside the object. One of the standard approaches for solving this problem is to reformulate it as an optimization problem: finding the boundary of the domain that minimizes the $L^2$ distance between computed values of the scattered field and the given measurement data. The optimization problem is computationally challenging since the local set of convexity shrinks with increasing frequency and results in an increasing number of local minima in the vicinity of the true solution. In many practical experimental settings, low frequency measurements are unavailable due to limitations of the experimental setup or the sensors used for measurement. Thus, obtaining a good initial guess for the optimization problem plays a vital role in this environment. We present a neural network warm-start approach for solving the inverse scattering problem, where an initial guess for the optimization problem is obtained using a trained neural network. We demonstrate the effectiveness of our method with several numerical examples. For high frequency problems, this approach outperforms traditional iterative methods such as Gauss-Newton initialized without any prior (i.e., initialized using a unit circle), or initialized using the solution of a direct method such as the linear sampling method. The algorithm remains robust to noise in the scattered field measurements and also converges to the true solution for limited aperture data. However, the number of training samples required to train the neural network scales exponentially in frequency and the complexity of the obstacles considered. We conclude with a discussion of this phenomenon and potential directions for future research.

Shape reconstruction of acoustic obstacle with linear sampling method and neural network

A Neural Network Warm-Start Approach for the Inverse Acoustic Obstacle Scattering Problem

Recovering a polyhedral obstacle by a few backscattering measurements

Proceedings of Meetings on Acoustics

A Robust Super-Resolution Approach with Sparsity Constraint in Acoustic Imaging

The inverse obstacle scattering with incident tapered waves

A Deep Learning Framework for Three Dimensional Shape Reconstruction from Phaseless Acoustic Scattering Far-field Data

Solving Inverse Obstacle Scattering Problem with Latent Surface Representations

A linear sampling method for inverse acoustic scattering by a locally rough interface

Solving Inverse Wave Scattering with Deep Learning

Inverse Obstacle scattering in two dimensions with multiple frequency data and multiple angles of incidence

Fast imaging of scattering obstacles from phaseless far-field measurements at a fixed frequency

Deep Neural Network-Oriented Indicator Method for Inverse Scattering Problems Using Partial Data

Reconstruction of inhomogeneous media by an iteration algorithm with a learned projector

Recovering scattering obstacles by multi-frequency phaseless far-field data

A Newton method for simultaneous reconstruction of an interface and a buried obstacle from far-field data

Inverse Obstacle Scattering in an Unbounded Structure

On an artificial neural network for inverse scattering problems

Mathematics-Informed Neural Network for Acoustic Scattering by Parallel Plates