Measuring representation

Lars Marstaller; Arend Hintze; Christoph Adami

doi:10.5096/ASCS200935

Measuring representation

Lars Marstaller, Arend Hintze, Christoph Adami

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

52 Downloads (Pure)

Abstract

We present a measure of representation in neural networks that we call ‘R’, which is based on information theory. We show how R relates to an analysis of distributed representation, viz. a principal components analysis of activation space. Finally, we argue that R is well suited to measure representation in neural networks.

Original language	English
Title of host publication	ASCS09
Subtitle of host publication	proceedings of the 9th Conference of the Australasian Society for Cognitive Science
Editors	Wayne Christensen, Elizabeth Schier, John Sutton
Place of Publication	North Ryde, NSW
Publisher	Macquarie Centre for Cognitive Science
Pages	232-237
Number of pages	6
ISBN (Print)	9780646529189
DOIs	https://doi.org/10.5096/ASCS200935
Publication status	Published - 2010
Externally published	Yes
Event	Conference of the Australasian Society for Cognitive Science (9th : 2009) - Sydney Duration: 30 Sept 2009 → 2 Oct 2009

Conference

Conference	Conference of the Australasian Society for Cognitive Science (9th : 2009)
City	Sydney
Period	30/09/09 → 2/10/09

Bibliographical note

Copyright 2009 by the Australasian Society for Cognitive Science. Publisher version archived with the permission of the Editor, ASCS09 : Proceedings of the 9th Conference of the Australasian Society for Cognitive Science. This copy is available for individual, non-commercial use. Permission to reprint/republish this version for other uses must be obtained from the publisher.

Keywords

artificial neural networks
(distributed) representation
information theory
principal components analysis
measure

Access to Document

10.5096/ASCS200935

Publisher version (open access)Final published version, 236 KBLicence: Unspecified

Cite this

@inproceedings{e29dae092d1744908869e46642d6ec21,

title = "Measuring representation",

abstract = "We present a measure of representation in neural networks that we call {\textquoteleft}R{\textquoteright}, which is based on information theory. We show how R relates to an analysis of distributed representation, viz. a principal components analysis of activation space. Finally, we argue that R is well suited to measure representation in neural networks.",

keywords = "artificial neural networks, (distributed) representation, information theory, principal components analysis, measure",

author = "Lars Marstaller and Arend Hintze and Christoph Adami",

note = "Copyright 2009 by the Australasian Society for Cognitive Science. Publisher version archived with the permission of the Editor, ASCS09 : Proceedings of the 9th Conference of the Australasian Society for Cognitive Science. This copy is available for individual, non-commercial use. Permission to reprint/republish this version for other uses must be obtained from the publisher.; Conference of the Australasian Society for Cognitive Science (9th : 2009) ; Conference date: 30-09-2009 Through 02-10-2009",

year = "2010",

doi = "10.5096/ASCS200935",

language = "English",

isbn = "9780646529189",

pages = "232--237",

editor = "Wayne Christensen and Elizabeth Schier and John Sutton",

booktitle = "ASCS09",

publisher = "Macquarie Centre for Cognitive Science",

}

Marstaller, L, Hintze, A & Adami, C 2010, Measuring representation. in W Christensen, E Schier & J Sutton (eds), ASCS09: proceedings of the 9th Conference of the Australasian Society for Cognitive Science. Macquarie Centre for Cognitive Science, North Ryde, NSW, pp. 232-237, Conference of the Australasian Society for Cognitive Science (9th : 2009), Sydney, 30/09/09. https://doi.org/10.5096/ASCS200935

Measuring representation. / Marstaller, Lars; Hintze, Arend; Adami, Christoph.
ASCS09: proceedings of the 9th Conference of the Australasian Society for Cognitive Science. ed. / Wayne Christensen; Elizabeth Schier; John Sutton. North Ryde, NSW: Macquarie Centre for Cognitive Science, 2010. p. 232-237.

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

TY - GEN

T1 - Measuring representation

AU - Marstaller, Lars

AU - Hintze, Arend

AU - Adami, Christoph

N1 - Copyright 2009 by the Australasian Society for Cognitive Science. Publisher version archived with the permission of the Editor, ASCS09 : Proceedings of the 9th Conference of the Australasian Society for Cognitive Science. This copy is available for individual, non-commercial use. Permission to reprint/republish this version for other uses must be obtained from the publisher.

PY - 2010

Y1 - 2010

N2 - We present a measure of representation in neural networks that we call ‘R’, which is based on information theory. We show how R relates to an analysis of distributed representation, viz. a principal components analysis of activation space. Finally, we argue that R is well suited to measure representation in neural networks.

AB - We present a measure of representation in neural networks that we call ‘R’, which is based on information theory. We show how R relates to an analysis of distributed representation, viz. a principal components analysis of activation space. Finally, we argue that R is well suited to measure representation in neural networks.

KW - artificial neural networks

KW - (distributed) representation

KW - information theory

KW - principal components analysis

KW - measure

U2 - 10.5096/ASCS200935

DO - 10.5096/ASCS200935

M3 - Conference proceeding contribution

SN - 9780646529189

SP - 232

EP - 237

BT - ASCS09

A2 - Christensen, Wayne

A2 - Schier, Elizabeth

A2 - Sutton, John

PB - Macquarie Centre for Cognitive Science

CY - North Ryde, NSW

T2 - Conference of the Australasian Society for Cognitive Science (9th : 2009)

Y2 - 30 September 2009 through 2 October 2009

ER -

Measuring representation

Abstract

Conference

Bibliographical note

Keywords

Access to Document

Fingerprint

Cite this