Encoding structure in holographic reduced representations

Matthew Kelly, Dorothea Blostein, D. J.K. Mewhort

Research output: Contribution to journalArticle

10 Citations (Scopus)

Abstract

Vector Symbolic Architectures (VSAs) such as Holographic Reduced Representations (HRRs) are computational associative memories used by cognitive psychologists to model behavioural and neurological aspects of human memory. We present a novel analysis of the mathematics of VSAs and a novel technique for representing data in HRRs. Encoding and decoding in VSAs can be characterised by Latin squares. Successful encoding requires the structure of the data to be orthogonal to the structure of the Latin squares. However, HRRs can successfully encode vectors of locally structured data if vectors are shuffled. Shuffling results are illustrated using images but are applicable to any nonrandom data. The ability to use locally structured vectors provides a technique for detailed modelling of stimuli in HRR models.

Original languageEnglish (US)
Pages (from-to)79-93
Number of pages15
JournalCanadian Journal of Experimental Psychology
Volume67
Issue number2
DOIs
StatePublished - Jun 1 2013

Fingerprint

Neurological Models
Aptitude
Mathematics
Psychology

All Science Journal Classification (ASJC) codes

  • Experimental and Cognitive Psychology

Cite this

Kelly, Matthew ; Blostein, Dorothea ; Mewhort, D. J.K. / Encoding structure in holographic reduced representations. In: Canadian Journal of Experimental Psychology. 2013 ; Vol. 67, No. 2. pp. 79-93.
@article{f62c3e728a8845f7bd51408329b3e7c4,
title = "Encoding structure in holographic reduced representations",
abstract = "Vector Symbolic Architectures (VSAs) such as Holographic Reduced Representations (HRRs) are computational associative memories used by cognitive psychologists to model behavioural and neurological aspects of human memory. We present a novel analysis of the mathematics of VSAs and a novel technique for representing data in HRRs. Encoding and decoding in VSAs can be characterised by Latin squares. Successful encoding requires the structure of the data to be orthogonal to the structure of the Latin squares. However, HRRs can successfully encode vectors of locally structured data if vectors are shuffled. Shuffling results are illustrated using images but are applicable to any nonrandom data. The ability to use locally structured vectors provides a technique for detailed modelling of stimuli in HRR models.",
author = "Matthew Kelly and Dorothea Blostein and Mewhort, {D. J.K.}",
year = "2013",
month = "6",
day = "1",
doi = "10.1037/a0030301",
language = "English (US)",
volume = "67",
pages = "79--93",
journal = "Canadian Journal of Experimental Psychology",
issn = "1196-1961",
publisher = "Canadian Psychological Association",
number = "2",

}

Encoding structure in holographic reduced representations. / Kelly, Matthew; Blostein, Dorothea; Mewhort, D. J.K.

In: Canadian Journal of Experimental Psychology, Vol. 67, No. 2, 01.06.2013, p. 79-93.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Encoding structure in holographic reduced representations

AU - Kelly, Matthew

AU - Blostein, Dorothea

AU - Mewhort, D. J.K.

PY - 2013/6/1

Y1 - 2013/6/1

N2 - Vector Symbolic Architectures (VSAs) such as Holographic Reduced Representations (HRRs) are computational associative memories used by cognitive psychologists to model behavioural and neurological aspects of human memory. We present a novel analysis of the mathematics of VSAs and a novel technique for representing data in HRRs. Encoding and decoding in VSAs can be characterised by Latin squares. Successful encoding requires the structure of the data to be orthogonal to the structure of the Latin squares. However, HRRs can successfully encode vectors of locally structured data if vectors are shuffled. Shuffling results are illustrated using images but are applicable to any nonrandom data. The ability to use locally structured vectors provides a technique for detailed modelling of stimuli in HRR models.

AB - Vector Symbolic Architectures (VSAs) such as Holographic Reduced Representations (HRRs) are computational associative memories used by cognitive psychologists to model behavioural and neurological aspects of human memory. We present a novel analysis of the mathematics of VSAs and a novel technique for representing data in HRRs. Encoding and decoding in VSAs can be characterised by Latin squares. Successful encoding requires the structure of the data to be orthogonal to the structure of the Latin squares. However, HRRs can successfully encode vectors of locally structured data if vectors are shuffled. Shuffling results are illustrated using images but are applicable to any nonrandom data. The ability to use locally structured vectors provides a technique for detailed modelling of stimuli in HRR models.

UR - http://www.scopus.com/inward/record.url?scp=84884555813&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84884555813&partnerID=8YFLogxK

U2 - 10.1037/a0030301

DO - 10.1037/a0030301

M3 - Article

C2 - 23205508

AN - SCOPUS:84884555813

VL - 67

SP - 79

EP - 93

JO - Canadian Journal of Experimental Psychology

JF - Canadian Journal of Experimental Psychology

SN - 1196-1961

IS - 2

ER -