Performance-enhancing bifurcations in a self-organising neural network

Terence Kwok; Kate Amanda Smith

doi:10.1007/3-540-44868-3_50

Performance-enhancing bifurcations in a self-organising neural network

Terence Kwok, Kate Amanda Smith

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

4 Citations (Scopus)

Abstract

The self-organising neural network with weight normalisation (SONN-WN) for solving combinatorial optimisation problems (COPs) is investigated in terms of its performance and dynamical characteristics. A simplified computational model of the weight normalisation process is constructed, which reveals symmetry-breaking bifurcations in a typical node outside the winning neighbourhood. Experimental results with the N-queen problem show that bifurcations can enhance solution qualities in a consistent manner. A mechanism based on the weights’ transient trajectories is proposed to account for the neural network’s capacity to escape local minima.

Original language	English
Title of host publication	Computational Methods in Neural Modeling
Subtitle of host publication	7th International Work-Conference on Artificial and Natural Neural Networks, IWANN 2003, Maó, Menorca, Spain, June 3-6. Proceedings, Part I
Editors	Jose Mira, Jose R Alvarez
Place of Publication	New York NY USA
Publisher	Springer
Pages	390-397
Number of pages	8
ISBN (Electronic)	9783540448686
ISBN (Print)	9783540402107
DOIs	https://doi.org/10.1007/3-540-44868-3_50
Publication status	Published - 2003
Event	International Work-Conference on Artificial and Natural Neural Networks 2003 - Mao Menorca Spain, New York USA Duration: 1 Jan 2003 → …

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
Volume	2686
ISSN (Print)	0302-9743

Conference

Conference	International Work-Conference on Artificial and Natural Neural Networks 2003
City	New York USA
Period	1/01/03 → …

Access to Document

10.1007/3-540-44868-3_50

Cite this

Kwok, T., & Smith, K. A. (2003). Performance-enhancing bifurcations in a self-organising neural network. In J. Mira, & J. R. Alvarez (Eds.), Computational Methods in Neural Modeling: 7th International Work-Conference on Artificial and Natural Neural Networks, IWANN 2003, Maó, Menorca, Spain, June 3-6. Proceedings, Part I (pp. 390-397). (Lecture Notes in Computer Science; Vol. 2686). Springer. https://doi.org/10.1007/3-540-44868-3_50

Kwok, Terence ; Smith, Kate Amanda. / Performance-enhancing bifurcations in a self-organising neural network. Computational Methods in Neural Modeling: 7th International Work-Conference on Artificial and Natural Neural Networks, IWANN 2003, Maó, Menorca, Spain, June 3-6. Proceedings, Part I. editor / Jose Mira ; Jose R Alvarez. New York NY USA : Springer, 2003. pp. 390-397 (Lecture Notes in Computer Science).

@inproceedings{17e8cdd9f5c247e4b1b5da5027af7a85,

title = "Performance-enhancing bifurcations in a self-organising neural network",

abstract = "The self-organising neural network with weight normalisation (SONN-WN) for solving combinatorial optimisation problems (COPs) is investigated in terms of its performance and dynamical characteristics. A simplified computational model of the weight normalisation process is constructed, which reveals symmetry-breaking bifurcations in a typical node outside the winning neighbourhood. Experimental results with the N-queen problem show that bifurcations can enhance solution qualities in a consistent manner. A mechanism based on the weights{\textquoteright} transient trajectories is proposed to account for the neural network{\textquoteright}s capacity to escape local minima.",

author = "Terence Kwok and Smith, \{Kate Amanda\}",

year = "2003",

doi = "10.1007/3-540-44868-3\_50",

language = "English",

isbn = "9783540402107",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "390--397",

editor = "Jose Mira and Alvarez, \{Jose R\}",

booktitle = "Computational Methods in Neural Modeling",

address = "Switzerland",

note = "International Work-Conference on Artificial and Natural Neural Networks 2003 ; Conference date: 01-01-2003",

}

Kwok, T & Smith, KA 2003, Performance-enhancing bifurcations in a self-organising neural network. in J Mira & JR Alvarez (eds), Computational Methods in Neural Modeling: 7th International Work-Conference on Artificial and Natural Neural Networks, IWANN 2003, Maó, Menorca, Spain, June 3-6. Proceedings, Part I. Lecture Notes in Computer Science, vol. 2686, Springer, New York NY USA, pp. 390-397, International Work-Conference on Artificial and Natural Neural Networks 2003, New York USA, 1/01/03. https://doi.org/10.1007/3-540-44868-3_50

Performance-enhancing bifurcations in a self-organising neural network. / Kwok, Terence; Smith, Kate Amanda.
Computational Methods in Neural Modeling: 7th International Work-Conference on Artificial and Natural Neural Networks, IWANN 2003, Maó, Menorca, Spain, June 3-6. Proceedings, Part I. ed. / Jose Mira; Jose R Alvarez. New York NY USA: Springer, 2003. p. 390-397 (Lecture Notes in Computer Science; Vol. 2686).

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

TY - GEN

T1 - Performance-enhancing bifurcations in a self-organising neural network

AU - Kwok, Terence

AU - Smith, Kate Amanda

PY - 2003

Y1 - 2003

N2 - The self-organising neural network with weight normalisation (SONN-WN) for solving combinatorial optimisation problems (COPs) is investigated in terms of its performance and dynamical characteristics. A simplified computational model of the weight normalisation process is constructed, which reveals symmetry-breaking bifurcations in a typical node outside the winning neighbourhood. Experimental results with the N-queen problem show that bifurcations can enhance solution qualities in a consistent manner. A mechanism based on the weights’ transient trajectories is proposed to account for the neural network’s capacity to escape local minima.

AB - The self-organising neural network with weight normalisation (SONN-WN) for solving combinatorial optimisation problems (COPs) is investigated in terms of its performance and dynamical characteristics. A simplified computational model of the weight normalisation process is constructed, which reveals symmetry-breaking bifurcations in a typical node outside the winning neighbourhood. Experimental results with the N-queen problem show that bifurcations can enhance solution qualities in a consistent manner. A mechanism based on the weights’ transient trajectories is proposed to account for the neural network’s capacity to escape local minima.

U2 - 10.1007/3-540-44868-3_50

DO - 10.1007/3-540-44868-3_50

M3 - Conference Paper

SN - 9783540402107

T3 - Lecture Notes in Computer Science

SP - 390

EP - 397

BT - Computational Methods in Neural Modeling

A2 - Mira, Jose

A2 - Alvarez, Jose R

PB - Springer

CY - New York NY USA

T2 - International Work-Conference on Artificial and Natural Neural Networks 2003

Y2 - 1 January 2003

ER -

Kwok T, Smith KA. Performance-enhancing bifurcations in a self-organising neural network. In Mira J, Alvarez JR, editors, Computational Methods in Neural Modeling: 7th International Work-Conference on Artificial and Natural Neural Networks, IWANN 2003, Maó, Menorca, Spain, June 3-6. Proceedings, Part I. New York NY USA: Springer. 2003. p. 390-397. (Lecture Notes in Computer Science). doi: 10.1007/3-540-44868-3_50