Optimization of multilayer photonic structures using artificial neural networks to obtain a target optical response

A new deep machine learning method is proposed for the task of selecting the parameters of a multilayer photonic structure to obtain a target optical spectrum of the reflection coefficient. The proposed training method is based on the connection of an artificial neural network for solving the inverse problem and the analytical transfer matrix method. This approach allows achieving high accuracy of the network. The developed method can be applied to the design of a structure that takes the derivative of the coordinate for an incident optical signal.