Studies on General Deep Neural Networks to Predict Optical Responses of Multilayered Thin Films

Abstract

Burgeoning demand on the analysis of multilayered optical thin films yields an advent of the simulation methods to calculate the optical response of structures. Machine learning ensures expedite calculation once a model has been trained. Here, we present a basic structure of deep neural network to demonstrate a validity of machine learning-based approach by showing how well the trained models approximate the target optical spectral responses. The numerical simulations with several tunable conditions show the overall inclination of performance of the models. This study paves the way for using machine learning-based approaches to predict the spectral responses of optical thin films.