Design and analysis of fixed planar holographic interconnects for optical neural networks

Paul E. Keller, Arthur F Gmitro

Research output: Contribution to journalArticle

20 Citations (Scopus)

Abstract

A neural-network architecture of multifaceted planar interconnection holograms and optoelectronic neurons is analyzed. Various computer-generated hologram techniques are analyzed and tested for their ability to produce an interconnection hologram with high-accuracy interconnects and high diffraction efficiency. A new technique is developed by using the Gerchberg-Saxton algorithm, followed by a random-search error minimization that produces the highest interconnect accuracy and the highest diffraction efficiency of the techniques tested. Analysis of the system shows that the hologram has the capacity to connect 5000 neuron outputs to 5000 neuron inputs with bipolar synapses and that the encoded synaptic weights have an accuracy of ˜ 5 bits. A simple feedback system is constructed and demonstrated.

Original languageEnglish (US)
Pages (from-to)5517-5526
Number of pages10
JournalApplied Optics
Volume31
Issue number26
DOIs
StatePublished - 1992

Fingerprint

Holograms
neurons
Neurons
Neural networks
Diffraction efficiency
synapses
diffraction
Network architecture
Optoelectronic devices
optimization
output
Feedback

Keywords

  • Computer-generated holograms
  • Multifaceted holograms
  • Neural networks
  • Optical computing
  • Optical interconnects

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics

Cite this

Design and analysis of fixed planar holographic interconnects for optical neural networks. / Keller, Paul E.; Gmitro, Arthur F.

In: Applied Optics, Vol. 31, No. 26, 1992, p. 5517-5526.

Research output: Contribution to journalArticle

@article{a43830c7eb2340ed9be3036483ce4331,
title = "Design and analysis of fixed planar holographic interconnects for optical neural networks",
abstract = "A neural-network architecture of multifaceted planar interconnection holograms and optoelectronic neurons is analyzed. Various computer-generated hologram techniques are analyzed and tested for their ability to produce an interconnection hologram with high-accuracy interconnects and high diffraction efficiency. A new technique is developed by using the Gerchberg-Saxton algorithm, followed by a random-search error minimization that produces the highest interconnect accuracy and the highest diffraction efficiency of the techniques tested. Analysis of the system shows that the hologram has the capacity to connect 5000 neuron outputs to 5000 neuron inputs with bipolar synapses and that the encoded synaptic weights have an accuracy of ˜ 5 bits. A simple feedback system is constructed and demonstrated.",
keywords = "Computer-generated holograms, Multifaceted holograms, Neural networks, Optical computing, Optical interconnects",
author = "Keller, {Paul E.} and Gmitro, {Arthur F}",
year = "1992",
doi = "10.1364/AO.31.005517",
language = "English (US)",
volume = "31",
pages = "5517--5526",
journal = "Applied Optics",
issn = "1559-128X",
publisher = "The Optical Society",
number = "26",

}

TY - JOUR

T1 - Design and analysis of fixed planar holographic interconnects for optical neural networks

AU - Keller, Paul E.

AU - Gmitro, Arthur F

PY - 1992

Y1 - 1992

N2 - A neural-network architecture of multifaceted planar interconnection holograms and optoelectronic neurons is analyzed. Various computer-generated hologram techniques are analyzed and tested for their ability to produce an interconnection hologram with high-accuracy interconnects and high diffraction efficiency. A new technique is developed by using the Gerchberg-Saxton algorithm, followed by a random-search error minimization that produces the highest interconnect accuracy and the highest diffraction efficiency of the techniques tested. Analysis of the system shows that the hologram has the capacity to connect 5000 neuron outputs to 5000 neuron inputs with bipolar synapses and that the encoded synaptic weights have an accuracy of ˜ 5 bits. A simple feedback system is constructed and demonstrated.

AB - A neural-network architecture of multifaceted planar interconnection holograms and optoelectronic neurons is analyzed. Various computer-generated hologram techniques are analyzed and tested for their ability to produce an interconnection hologram with high-accuracy interconnects and high diffraction efficiency. A new technique is developed by using the Gerchberg-Saxton algorithm, followed by a random-search error minimization that produces the highest interconnect accuracy and the highest diffraction efficiency of the techniques tested. Analysis of the system shows that the hologram has the capacity to connect 5000 neuron outputs to 5000 neuron inputs with bipolar synapses and that the encoded synaptic weights have an accuracy of ˜ 5 bits. A simple feedback system is constructed and demonstrated.

KW - Computer-generated holograms

KW - Multifaceted holograms

KW - Neural networks

KW - Optical computing

KW - Optical interconnects

UR - http://www.scopus.com/inward/record.url?scp=0010928543&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0010928543&partnerID=8YFLogxK

U2 - 10.1364/AO.31.005517

DO - 10.1364/AO.31.005517

M3 - Article

AN - SCOPUS:0010928543

VL - 31

SP - 5517

EP - 5526

JO - Applied Optics

JF - Applied Optics

SN - 1559-128X

IS - 26

ER -