Machine learning has been received considerable attention in various research fields, because the processing power of general-purpose computers has improved. The design of optical materials using neural network (NN) may revolutionize the methodology of nanophotonics research. In this paper, we introduce NN methods to solve the problems of scattering cross-section of multi-layered cylinder (forward calculation). The NN calculation is 3500 times faster than that based on the Mie theory.
Paper
Full text
Design of metamaterials using neural networks
Semantic Scholar · Physics · 2019
Abstract
Machine learning has been received considerable attention in various research fields, because the processing power of general-purpose computers has improved. The design of optical materials using neural network (NN) may revolutionize the methodology of nanophotonics research. In this paper, we introduce NN methods to solve the problems of scattering cross-section of multi-layered cylinder (forward calculation). The NN calculation is 3500 times faster than that based on the Mie theory.