Electromagnetic Absorbing Materials Design by Optimization Using a Machine Learning Approach

One of the state-of-the-art strategies to decrease the isolation between sections of personal digital assistance devices, taking into account the unavoidable thermal, mechanical, weight, and manufacturability constraints is the proper introduction of electromagnetic (EM) absorbing materials. The global target of this study is the optimized synthesis of the EM properties of a material using a machine learning approach. The sought material must improve the shielding effectiveness between two regions of a device. The first part of this paper is dedicated to the assessment of the performances of an artificial neural network used as function approximation to compute the values of the shielding without resorting to complex and time-consuming full-wave numerical simulations. The network is assessed with respect to either the variations of the training set and algorithm or of its architecture. Significant figures of merit are introduced. The second part is devoted to the description and definition of nature-inspired algorithms demanded to manage the optimization problem.