A review of the most recent advances in deep learning (DL) as applied to electromagnetics (EM), antennas, and propagation is provided. It is aimed at giving the interested readers and pratictioners in EM and related applicative fields some useful insights on the effectiveness and the potentialities of deep neural networks (DNNs) as computational tools with an unprecedented computational efficiency. The range of considered applications includes forward/inverse scattering, direction-of-arrival (DoA) estimation, radar and remote sensing (RS), and multi-input/multi-output (MIMO) systems. Appealing DNN-based solutions concerned with localization, human behavior monitoring (HBM), and EM compatibility (EMC) are reported, as well. Some final remarks are drawn along with indications on future trends according to the authors' viewpoint.

An innovative inverse scattering (IS) technique for the simultaneous processing of multifrequency (MF) ground-penetrating radar (GPR) measurements is proposed. The nonlinear IS problem is solved by profitably integrating a customized MF version of the particle swarm optimizer (PSO) within the iterative multiscaling approach (IMSA) to jointly exploit the reduction of the ratio between unknowns and uncorrelated data with a pervasive exploration of the multidimensional search space for minimizing the probability that the solution is trapped into local minima corresponding to false solutions of the problem at hand. Both numerical and experimental test cases are reported to assess the reliability of the MF-IMSA-PSO method toward accurate GPR tomography as well as improvements with respect to the competitive state-of-the-art inversion approaches.

An innovative approachfor the synthesis of inexpensive holographic smart electromagnetic (EM) skins with advanced beamforming features is proposed. The complex multi-scale smart skin design is formulated within the Generalized Sheet Transition Condition (GSTC) framework as a combination of a mask-constrained isophoric inverse source problem and a micro-scale susceptibility dyadic optimization. The solution strategy integrates a local search procedure based on the iterative projection technique (IPT) and a System-by-Design(SbD)-based optimization loop for the identification of optimal metasurface descriptors matching the desired surface currents. The performance and the efficiency of the proposed approach are assessed in a set of representative numerical test cases concerned with different smart skin apertures and target pattern masks.

In recent years, artificial intelligence (AI) techniques have been developed rapidly. With the help of big data, massive parallel computing, and optimization algorithms, machine learning (ML) and (more recently) deep learning (DL) strategies have been equipped with enhanced learning and generalization capabilities. Besides becoming an essential framework in image and speech signal processing, AI has also been widely applied to solve several electromagnetic (EM) problems with unprecedented computational efficiency, including inverse scattering (IS) and EM imaging. In this article, a review of the most recent progresses in the application of ML and DL for such problems is given. We humbly hope a brief summary could help us better understand the pros and cons of this research topic and foster future research in using AI to address paramount challenges in the field of EM vision.

The application of the compressive sensing (CS) paradigm to retrieve non-single-pixels contrast profiles is discussed. By exploiting a wavelet representation to model complex scatterer distributions with sparse vectors of coefficients, an efficient Bayesian CS (BCS) strategy is adopted to solve the arising inverse scattering problem. A set of representative numerical examples is presented to illustrate the advantages and the limitations of the proposed approach also with respect to comparable state-of-the-art inversion methods.

The design of a smart electromagnetic (EM) environment for next-generation wireless communication systems is addressed in this work. The proposed approach aims at synthesizing a desired EM field distribution over a target region, where the receiving terminals are located, through the opportunistic exploitation of the complex scattering interactions between the EM field generated by a reconfigurable primary source and the objects/scatterers present in the environment, which behave as application-driven passive metastructures. The effectiveness and the potentialities of the proposed design methodology are assessed with a proof-of-concept numerical result obtained by means of advanced and reliable simulation tools.