Deep learning enabled invisibility cloak design

Abstract

Invisibility cloak has become a hot topic recently, and development has been working on it. In this project, approach is made to reduce scattering of invisibility cloak using optimization and deep leaning techniques. Two different approaches were experimented in this project, first one was using Optimization Toolbox optimized parameters to train a MLP model. The second one was by using Global Optimization Toolbox’s MultiStart solver to optimize parameters. After analyzing, the second one yields a better performance which resulted in lower scattering energy, with the only downside of longer computation time.