Comparison of Models for Training Optical Matrix Multipliers in Neuromorphic PICs

Ali Cem*, Siqi Yan, Uiara Celine de Moura, Yunhong Ding, Darko Zibar, Francesco Da Ros

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

129 Downloads (Pure)

Abstract

We experimentally compare simple physics-based vs. data-driven neural-network-based models for offline training of programmable photonic chips using Mach-Zehnder interferometer meshes. The neural-network model outperforms physics-based models for a chip with thermal crosstalk, yielding increased testing accuracy.
Original languageEnglish
Title of host publicationProceedings of OFC 2022
Number of pages3
PublisherOptical Society of America (OSA)
Publication date2022
Article numberM2G.5
Publication statusPublished - 2022
Event2022 Optical Fiber Communications Conference and Exhibition - San Diego Convention Center, San Diego, United States
Duration: 6 Mar 202210 Mar 2022

Conference

Conference2022 Optical Fiber Communications Conference and Exhibition
LocationSan Diego Convention Center
Country/TerritoryUnited States
CitySan Diego
Period06/03/202210/03/2022

Fingerprint

Dive into the research topics of 'Comparison of Models for Training Optical Matrix Multipliers in Neuromorphic PICs'. Together they form a unique fingerprint.

Cite this