γ-Observable neighbours for vector quantization

Michael Aupetit, Pierre Couturier, Pierre Massotte

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

We define the γ-observable neighbourhood and use it in soft-competitive learning for vector quantization. Considering a datum v and a set of n units wi in a Euclidean space, let vi be a point of the segment [vwi] whose position depends on γ a real number between 0 and 1, the γ-observable neighbours (γ-ON) of v are the units wi for which vi is in the Voronoï of wi, i.e. wi is the closest unit to vi. For γ=1, vi merges with wi, all the units are γ-ON of v, while for γ=0, vi merges with v, only the closest unit to v is its γ-ON. The size of the neighbourhood decreases from n to 1 while γ goes from 1 to 0. For γ lower or equal to 0.5, the γ-ON of v are also its natural neighbours, i.e. their Voronoï regions share a common boundary with that of v. We show that this neighbourhood used in Vector Quantization gives faster convergence in terms of number of epochs and similar distortion than the Neural-Gas on several benchmark databases, and we propose the fact that it does not have the dimension selection property could explain these results. We show it also presents a new self-organization property we call 'self-distribution'.

Original languageEnglish
Pages (from-to)1017-1027
Number of pages11
JournalNeural Networks
Volume15
Issue number8-9
DOIs
Publication statusPublished - Oct 2002
Externally publishedYes

Fingerprint

Vector quantization
Benchmarking
Gases
Learning
Databases

Keywords

  • γ-Observable neighbours
  • Dimension selection
  • Natural neighbours
  • Neural-gas
  • Self-distribution
  • Self-organizing maps
  • Vector quantization

ASJC Scopus subject areas

  • Artificial Intelligence
  • Neuroscience(all)

Cite this

γ-Observable neighbours for vector quantization. / Aupetit, Michael; Couturier, Pierre; Massotte, Pierre.

In: Neural Networks, Vol. 15, No. 8-9, 10.2002, p. 1017-1027.

Research output: Contribution to journalArticle

Aupetit, Michael ; Couturier, Pierre ; Massotte, Pierre. / γ-Observable neighbours for vector quantization. In: Neural Networks. 2002 ; Vol. 15, No. 8-9. pp. 1017-1027.
@article{f8a3b68169c041daae35d42a1c2716b0,
title = "γ-Observable neighbours for vector quantization",
abstract = "We define the γ-observable neighbourhood and use it in soft-competitive learning for vector quantization. Considering a datum v and a set of n units wi in a Euclidean space, let vi be a point of the segment [vwi] whose position depends on γ a real number between 0 and 1, the γ-observable neighbours (γ-ON) of v are the units wi for which vi is in the Vorono{\"i} of wi, i.e. wi is the closest unit to vi. For γ=1, vi merges with wi, all the units are γ-ON of v, while for γ=0, vi merges with v, only the closest unit to v is its γ-ON. The size of the neighbourhood decreases from n to 1 while γ goes from 1 to 0. For γ lower or equal to 0.5, the γ-ON of v are also its natural neighbours, i.e. their Vorono{\"i} regions share a common boundary with that of v. We show that this neighbourhood used in Vector Quantization gives faster convergence in terms of number of epochs and similar distortion than the Neural-Gas on several benchmark databases, and we propose the fact that it does not have the dimension selection property could explain these results. We show it also presents a new self-organization property we call 'self-distribution'.",
keywords = "γ-Observable neighbours, Dimension selection, Natural neighbours, Neural-gas, Self-distribution, Self-organizing maps, Vector quantization",
author = "Michael Aupetit and Pierre Couturier and Pierre Massotte",
year = "2002",
month = "10",
doi = "10.1016/S0893-6080(02)00076-X",
language = "English",
volume = "15",
pages = "1017--1027",
journal = "Neural Networks",
issn = "0893-6080",
publisher = "Elsevier Limited",
number = "8-9",

}

TY - JOUR

T1 - γ-Observable neighbours for vector quantization

AU - Aupetit, Michael

AU - Couturier, Pierre

AU - Massotte, Pierre

PY - 2002/10

Y1 - 2002/10

N2 - We define the γ-observable neighbourhood and use it in soft-competitive learning for vector quantization. Considering a datum v and a set of n units wi in a Euclidean space, let vi be a point of the segment [vwi] whose position depends on γ a real number between 0 and 1, the γ-observable neighbours (γ-ON) of v are the units wi for which vi is in the Voronoï of wi, i.e. wi is the closest unit to vi. For γ=1, vi merges with wi, all the units are γ-ON of v, while for γ=0, vi merges with v, only the closest unit to v is its γ-ON. The size of the neighbourhood decreases from n to 1 while γ goes from 1 to 0. For γ lower or equal to 0.5, the γ-ON of v are also its natural neighbours, i.e. their Voronoï regions share a common boundary with that of v. We show that this neighbourhood used in Vector Quantization gives faster convergence in terms of number of epochs and similar distortion than the Neural-Gas on several benchmark databases, and we propose the fact that it does not have the dimension selection property could explain these results. We show it also presents a new self-organization property we call 'self-distribution'.

AB - We define the γ-observable neighbourhood and use it in soft-competitive learning for vector quantization. Considering a datum v and a set of n units wi in a Euclidean space, let vi be a point of the segment [vwi] whose position depends on γ a real number between 0 and 1, the γ-observable neighbours (γ-ON) of v are the units wi for which vi is in the Voronoï of wi, i.e. wi is the closest unit to vi. For γ=1, vi merges with wi, all the units are γ-ON of v, while for γ=0, vi merges with v, only the closest unit to v is its γ-ON. The size of the neighbourhood decreases from n to 1 while γ goes from 1 to 0. For γ lower or equal to 0.5, the γ-ON of v are also its natural neighbours, i.e. their Voronoï regions share a common boundary with that of v. We show that this neighbourhood used in Vector Quantization gives faster convergence in terms of number of epochs and similar distortion than the Neural-Gas on several benchmark databases, and we propose the fact that it does not have the dimension selection property could explain these results. We show it also presents a new self-organization property we call 'self-distribution'.

KW - γ-Observable neighbours

KW - Dimension selection

KW - Natural neighbours

KW - Neural-gas

KW - Self-distribution

KW - Self-organizing maps

KW - Vector quantization

UR - http://www.scopus.com/inward/record.url?scp=0036789801&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0036789801&partnerID=8YFLogxK

U2 - 10.1016/S0893-6080(02)00076-X

DO - 10.1016/S0893-6080(02)00076-X

M3 - Article

VL - 15

SP - 1017

EP - 1027

JO - Neural Networks

JF - Neural Networks

SN - 0893-6080

IS - 8-9

ER -