Electrooptic nonlinear activation functions for vector matrix multiplications in optical neural networks

Jonathan George, Rubab Amin, Armin Mehrabian, Jacob Khurgin, Tarek El-Ghazawi, Paul R. Prucnal, Volker J. Sorger

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations

Abstract

We show how the nonlinear transfer function of electrooptic modulators enables vector matrix multiplications of photonic neural networks. Here the modulators energy-per-bit function and signal-to-noise ratio are critical factors impacting system performance.

Original languageEnglish (US)
Title of host publicationSignal Processing in Photonic Communications, SPPCom 2018
PublisherOSA - The Optical Society
ISBN (Electronic)9781557528209
DOIs
StatePublished - Jan 1 2018
EventSignal Processing in Photonic Communications, SPPCom 2018 - Zurich, Switzerland
Duration: Jul 2 2018Jul 5 2018

Publication series

NameOptics InfoBase Conference Papers
VolumePart F112-SPPCom 2018

Other

OtherSignal Processing in Photonic Communications, SPPCom 2018
CountrySwitzerland
CityZurich
Period7/2/187/5/18

All Science Journal Classification (ASJC) codes

  • Electronic, Optical and Magnetic Materials
  • Mechanics of Materials

Fingerprint Dive into the research topics of 'Electrooptic nonlinear activation functions for vector matrix multiplications in optical neural networks'. Together they form a unique fingerprint.

  • Cite this

    George, J., Amin, R., Mehrabian, A., Khurgin, J., El-Ghazawi, T., Prucnal, P. R., & Sorger, V. J. (2018). Electrooptic nonlinear activation functions for vector matrix multiplications in optical neural networks. In Signal Processing in Photonic Communications, SPPCom 2018 (Optics InfoBase Conference Papers; Vol. Part F112-SPPCom 2018). OSA - The Optical Society. https://doi.org/10.1364/SPPCom.2018.SpW4G.3