Parametric sensitivity and scalability of k-winners-take-all networks with a single state variable and infinity-gain activation functions

Jun Wang; Zhishan Guo

doi:10.1007/978-3-642-13278-0_11

Parametric sensitivity and scalability of k-winners-take-all networks with a single state variable and infinity-gain activation functions

Jun Wang
, Zhishan Guo

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

15 Citations (Scopus)

Abstract

In recent years, several k-winners-take-all (kWTA) neural networks were developed based on a quadratic programming formulation. In particular, a continuous-time kWTA network with a single state variable and its discrete-time counterpart were developed recently. These kWTA networks have proven properties of global convergence and simple architectures. Starting with problem formulations, this paper reviews related existing kWTA networks and extends the existing kWTA networks with piecewise linear activation functions to the ones with high-gain activation functions. The paper then presents experimental results of the continuous-time and discrete-time kWTA networks with infinity-gain activation functions. The results show that the kWTA networks are parametrically robust and dimensionally scalable in terms of problem size and convergence rate. © 2010 Springer-Verlag.

Original language	English
Title of host publication	Advances in Neural Networks - ISNN 2010
Subtitle of host publication	7th International Symposium on Neural Networks, ISNN 2010, Proceedings
Publisher	Springer Verlag
Pages	77-85
Volume	6063 LNCS
ISBN (Print)	3642132774, 9783642132773
DOIs	https://doi.org/10.1007/978-3-642-13278-0_11
Publication status	Published - 2010
Externally published	Yes
Event	7th International Symposium on Neural Networks, ISNN 2010 - Shanghai, China Duration: 6 Jun 2010 → 9 Jun 2010

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	6063 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	7th International Symposium on Neural Networks, ISNN 2010
Place	China
City	Shanghai
Period	6/06/10 → 9/06/10

Research Keywords

K winners-take-all
optimization
parametric sensitivity
recurrent neural networks
scalability

Access Output

10.1007/978-3-642-13278-0_11

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

Wang, J., & Guo, Z. (2010). Parametric sensitivity and scalability of k-winners-take-all networks with a single state variable and infinity-gain activation functions. In Advances in Neural Networks - ISNN 2010: 7th International Symposium on Neural Networks, ISNN 2010, Proceedings (Vol. 6063 LNCS, pp. 77-85). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6063 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-642-13278-0_11

Wang, Jun ; Guo, Zhishan. / Parametric sensitivity and scalability of k-winners-take-all networks with a single state variable and infinity-gain activation functions. Advances in Neural Networks - ISNN 2010: 7th International Symposium on Neural Networks, ISNN 2010, Proceedings. Vol. 6063 LNCS Springer Verlag, 2010. pp. 77-85 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{e40edd1c8b184504a651b471798c0ac5,

title = "Parametric sensitivity and scalability of k-winners-take-all networks with a single state variable and infinity-gain activation functions",

abstract = "In recent years, several k-winners-take-all (kWTA) neural networks were developed based on a quadratic programming formulation. In particular, a continuous-time kWTA network with a single state variable and its discrete-time counterpart were developed recently. These kWTA networks have proven properties of global convergence and simple architectures. Starting with problem formulations, this paper reviews related existing kWTA networks and extends the existing kWTA networks with piecewise linear activation functions to the ones with high-gain activation functions. The paper then presents experimental results of the continuous-time and discrete-time kWTA networks with infinity-gain activation functions. The results show that the kWTA networks are parametrically robust and dimensionally scalable in terms of problem size and convergence rate. {\textcopyright} 2010 Springer-Verlag.",

keywords = "K winners-take-all, optimization, parametric sensitivity, recurrent neural networks, scalability",

author = "Jun Wang and Zhishan Guo",

year = "2010",

doi = "10.1007/978-3-642-13278-0\_11",

language = "English",

isbn = "3642132774",

volume = "6063 LNCS",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "77--85",

booktitle = "Advances in Neural Networks - ISNN 2010",

address = "Germany",

note = "7th International Symposium on Neural Networks, ISNN 2010 ; Conference date: 06-06-2010 Through 09-06-2010",

}

Wang, J & Guo, Z 2010, Parametric sensitivity and scalability of k-winners-take-all networks with a single state variable and infinity-gain activation functions. in Advances in Neural Networks - ISNN 2010: 7th International Symposium on Neural Networks, ISNN 2010, Proceedings. vol. 6063 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6063 LNCS, Springer Verlag, pp. 77-85, 7th International Symposium on Neural Networks, ISNN 2010, Shanghai, China, 6/06/10. https://doi.org/10.1007/978-3-642-13278-0_11

Parametric sensitivity and scalability of k-winners-take-all networks with a single state variable and infinity-gain activation functions. / Wang, Jun; Guo, Zhishan.
Advances in Neural Networks - ISNN 2010: 7th International Symposium on Neural Networks, ISNN 2010, Proceedings. Vol. 6063 LNCS Springer Verlag, 2010. p. 77-85 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6063 LNCS).

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - Parametric sensitivity and scalability of k-winners-take-all networks with a single state variable and infinity-gain activation functions

AU - Wang, Jun

AU - Guo, Zhishan

PY - 2010

Y1 - 2010

N2 - In recent years, several k-winners-take-all (kWTA) neural networks were developed based on a quadratic programming formulation. In particular, a continuous-time kWTA network with a single state variable and its discrete-time counterpart were developed recently. These kWTA networks have proven properties of global convergence and simple architectures. Starting with problem formulations, this paper reviews related existing kWTA networks and extends the existing kWTA networks with piecewise linear activation functions to the ones with high-gain activation functions. The paper then presents experimental results of the continuous-time and discrete-time kWTA networks with infinity-gain activation functions. The results show that the kWTA networks are parametrically robust and dimensionally scalable in terms of problem size and convergence rate. © 2010 Springer-Verlag.

AB - In recent years, several k-winners-take-all (kWTA) neural networks were developed based on a quadratic programming formulation. In particular, a continuous-time kWTA network with a single state variable and its discrete-time counterpart were developed recently. These kWTA networks have proven properties of global convergence and simple architectures. Starting with problem formulations, this paper reviews related existing kWTA networks and extends the existing kWTA networks with piecewise linear activation functions to the ones with high-gain activation functions. The paper then presents experimental results of the continuous-time and discrete-time kWTA networks with infinity-gain activation functions. The results show that the kWTA networks are parametrically robust and dimensionally scalable in terms of problem size and convergence rate. © 2010 Springer-Verlag.

KW - K winners-take-all

KW - optimization

KW - parametric sensitivity

KW - recurrent neural networks

KW - scalability

UR - https://www.scopus.com/pages/publications/77954446618

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-77954446618&origin=recordpage

U2 - 10.1007/978-3-642-13278-0_11

DO - 10.1007/978-3-642-13278-0_11

M3 - RGC 32 - Refereed conference paper (with host publication)

SN - 3642132774

SN - 9783642132773

VL - 6063 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 77

EP - 85

BT - Advances in Neural Networks - ISNN 2010

PB - Springer Verlag

T2 - 7th International Symposium on Neural Networks, ISNN 2010

Y2 - 6 June 2010 through 9 June 2010

ER -

Wang J, Guo Z. Parametric sensitivity and scalability of k-winners-take-all networks with a single state variable and infinity-gain activation functions. In Advances in Neural Networks - ISNN 2010: 7th International Symposium on Neural Networks, ISNN 2010, Proceedings. Vol. 6063 LNCS. Springer Verlag. 2010. p. 77-85. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-642-13278-0_11

Parametric sensitivity and scalability of k-winners-take-all networks with a single state variable and infinity-gain activation functions

Abstract

Publication series

Conference

Research Keywords

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this