Learning-Based Procedures for Inverse Design of Electromagnetic Devices: A Preliminary Investigation

The concept of Inverse Design (ID) has grown in the last years in many branches of engineering. In this contribution, the possibility of performing optimal ID within the inverse scattering (IS) framework is presented. In particular, the possibility of exploiting deep learning (DL) based approaches is discussed by focusing on the joint exploitation of the most suitable DL architecture and IS techniques.