Predicting Effective Refractive Indices of Multimode Waveguide via Deep Learning

Abstract

In order to accelerate the multimode waveguide design, several regression models are employed to predict the effective refractive indices (neff) from fundamental mode to fourth-order TE mode with various waveguide geometric parameters. On dataset with air cladding, the percent of eligible data whose prediction error is less than 10–3 of different modes are respectively 89.95%, 88.10%, 82.29%, 75.83%, 71.19%. And on dataset with SiO2 cladding, they are 95.40%, 92.81 %, 90.90%, 81.99%, 86.39%. This study guides the structural design and optimization of optical waveguides based on machine learning.