Holographic Graph Neuron: A Bioinspired Architecture for Pattern Processing

Denis Kleyko, Evgeny Osipov, Alexander Senior, Asad I. Khan, Yasar Ahmet Sekercioglu

Research output: Contribution to journalArticle

Abstract

In this paper, we propose a new approach to implementing hierarchical graph neuron (HGN), an architecture for memorizing patterns of generic sensor stimuli, through the use of vector symbolic architectures. The adoption of a vector symbolic representation ensures a single-layer design while retaining the existing performance characteristics of HGN. This approach significantly improves the noise resistance of the HGN architecture, and enables a linear (with respect to the number of stored entries) time search for an arbitrary subpattern.

LanguageEnglish
Pages1250-1262
Number of pages13
JournalIEEE Transactions on Neural Networks and Learning Systems
Volume28
Issue number6
DOIs
StatePublished - Jun 2017

Keywords

  • Associative memory (AM)
  • Holographic graph neuron (HoloGN)
  • Hyperdimensional computing
  • Pattern recognition
  • Vector symbolic architectures (VSAs)

Cite this

Kleyko, Denis ; Osipov, Evgeny ; Senior, Alexander ; Khan, Asad I. ; Sekercioglu, Yasar Ahmet. / Holographic Graph Neuron : A Bioinspired Architecture for Pattern Processing. In: IEEE Transactions on Neural Networks and Learning Systems. 2017 ; Vol. 28, No. 6. pp. 1250-1262
@article{a42c5addd8f447ea93d5c000150ac354,
title = "Holographic Graph Neuron: A Bioinspired Architecture for Pattern Processing",
abstract = "In this paper, we propose a new approach to implementing hierarchical graph neuron (HGN), an architecture for memorizing patterns of generic sensor stimuli, through the use of vector symbolic architectures. The adoption of a vector symbolic representation ensures a single-layer design while retaining the existing performance characteristics of HGN. This approach significantly improves the noise resistance of the HGN architecture, and enables a linear (with respect to the number of stored entries) time search for an arbitrary subpattern.",
keywords = "Associative memory (AM), Holographic graph neuron (HoloGN), Hyperdimensional computing, Pattern recognition, Vector symbolic architectures (VSAs)",
author = "Denis Kleyko and Evgeny Osipov and Alexander Senior and Khan, {Asad I.} and Sekercioglu, {Yasar Ahmet}",
year = "2017",
month = "6",
doi = "10.1109/TNNLS.2016.2535338",
language = "English",
volume = "28",
pages = "1250--1262",
journal = "IEEE Transactions on Neural Networks and Learning Systems",
issn = "2162-237X",
number = "6",

}

Holographic Graph Neuron : A Bioinspired Architecture for Pattern Processing. / Kleyko, Denis; Osipov, Evgeny; Senior, Alexander ; Khan, Asad I.; Sekercioglu, Yasar Ahmet.

In: IEEE Transactions on Neural Networks and Learning Systems, Vol. 28, No. 6, 06.2017, p. 1250-1262.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Holographic Graph Neuron

T2 - IEEE Transactions on Neural Networks and Learning Systems

AU - Kleyko,Denis

AU - Osipov,Evgeny

AU - Senior,Alexander

AU - Khan,Asad I.

AU - Sekercioglu,Yasar Ahmet

PY - 2017/6

Y1 - 2017/6

N2 - In this paper, we propose a new approach to implementing hierarchical graph neuron (HGN), an architecture for memorizing patterns of generic sensor stimuli, through the use of vector symbolic architectures. The adoption of a vector symbolic representation ensures a single-layer design while retaining the existing performance characteristics of HGN. This approach significantly improves the noise resistance of the HGN architecture, and enables a linear (with respect to the number of stored entries) time search for an arbitrary subpattern.

AB - In this paper, we propose a new approach to implementing hierarchical graph neuron (HGN), an architecture for memorizing patterns of generic sensor stimuli, through the use of vector symbolic architectures. The adoption of a vector symbolic representation ensures a single-layer design while retaining the existing performance characteristics of HGN. This approach significantly improves the noise resistance of the HGN architecture, and enables a linear (with respect to the number of stored entries) time search for an arbitrary subpattern.

KW - Associative memory (AM)

KW - Holographic graph neuron (HoloGN)

KW - Hyperdimensional computing

KW - Pattern recognition

KW - Vector symbolic architectures (VSAs)

UR - http://www.scopus.com/inward/record.url?scp=84960540059&partnerID=8YFLogxK

U2 - 10.1109/TNNLS.2016.2535338

DO - 10.1109/TNNLS.2016.2535338

M3 - Article

VL - 28

SP - 1250

EP - 1262

JO - IEEE Transactions on Neural Networks and Learning Systems

JF - IEEE Transactions on Neural Networks and Learning Systems

SN - 2162-237X

IS - 6

ER -