Physics-guided Loss Functions Improve Deep Learning Performance in Inverse Scattering

TL;DR

This paper demonstrates that incorporating physics-based near-field priors into loss functions significantly enhances deep learning methods for electromagnetic inverse scattering, leading to improved imaging accuracy and robustness.

Contribution

The paper introduces novel physics-guided loss functions that embed near-field physical phenomena into deep neural network training for inverse scattering problems.

Findings

Physics-guided loss functions improve DNN imaging performance.

Including near-field priors enhances robustness against noise.

Analysis of different loss functions reveals their advantages and limitations.

Abstract

Solving electromagnetic inverse scattering problems (ISPs) is challenging due to the intrinsic nonlinearity, ill-posedness, and expensive computational cost. Recently, deep neural network (DNN) techniques have been successfully applied on ISPs and shown potential of superior imaging over conventional methods. In this paper, we analyse the analogy between DNN solvers and traditional iterative algorithms and discuss how important physical phenomena cannot be effectively incorporated in the training process. We show the importance of including near-field priors in the learning process of DNNs. To this end, we propose new designs of loss functions which incorporate multiple-scattering based near-field quantities (such as scattered fields or induced currents within domain of interest). Effects of physics-guided loss functions are studied using a variety of numerical experiments. Pros and cons of the investigated ISP solvers with different loss functions are summarized.