paper

Data-efficient Modeling of Optical Matrix Multipliers Using Transfer Learning

arXiv:2211.16038

Abstract

We demonstrate transfer learning-assisted neural network models for optical matrix multipliers with scarce measurement data. Our approach uses <10\% of experimental data needed for best performance and outperforms analytical models for a Mach-Zehnder interferometer mesh.

2 pages, 2 figues, submitted to CLEO

Data-efficient Modeling of Optical Matrix Multipliers Using Transfer Learning · wovepaper