Analysis on generalization error of faulty rbf networks with weight decay regularizer

Chi Sing Leung; Pui Fai Sum; Hongjiang Wang

doi:10.1007/978-3-642-03040-6_39

Analysis on generalization error of faulty rbf networks with weight decay regularizer

Chi Sing Leung
, Pui Fai Sum
, Hongjiang Wang

Department of Electrical Engineering

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

2 Citations (Scopus)

Abstract

In the past two decades, the use of the weight decay regularizer for improving the generalization ability of neural networks has been extensively investigated. However, most existing results focus on the fault-free neural networks only. This papers extends the analysis on the generalization ability for networks with multiplicative weight noise. Our analysis result allows us not only to estimate the generalization ability of a faulty network, but also to select a good model from various settings. Simulated experiments are performed to verify theoretical result. © 2009 Springer Berlin Heidelberg.

Original language	English
Title of host publication	Advances in Neuro-Information Processing
Subtitle of host publication	15th International Conference, ICONIP 2008, Revised Selected Papers
Publisher	Springer Verlag
Pages	316-323
Volume	5507 LNCS
Edition	PART 2
ISBN (Print)	3642030394, 9783642030390
DOIs	https://doi.org/10.1007/978-3-642-03040-6_39
Publication status	Published - 2009
Event	15th International Conference on Neuro-Information Processing, ICONIP 2008 - Auckland, New Zealand Duration: 25 Nov 2008 → 28 Nov 2008

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	5507 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	15th International Conference on Neuro-Information Processing, ICONIP 2008
Place	New Zealand
City	Auckland
Period	25/11/08 → 28/11/08

Access Output

10.1007/978-3-642-03040-6_39

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

Leung, C. S., Sum, P. F., & Wang, H. (2009). Analysis on generalization error of faulty rbf networks with weight decay regularizer. In Advances in Neuro-Information Processing: 15th International Conference, ICONIP 2008, Revised Selected Papers (PART 2 ed., Vol. 5507 LNCS, pp. 316-323). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5507 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-642-03040-6_39

Leung, Chi Sing ; Sum, Pui Fai ; Wang, Hongjiang. / Analysis on generalization error of faulty rbf networks with weight decay regularizer. Advances in Neuro-Information Processing: 15th International Conference, ICONIP 2008, Revised Selected Papers. Vol. 5507 LNCS PART 2. ed. Springer Verlag, 2009. pp. 316-323 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{8e7f9eacb67a4762af1dbb53de4f8e4a,

title = "Analysis on generalization error of faulty rbf networks with weight decay regularizer",

abstract = "In the past two decades, the use of the weight decay regularizer for improving the generalization ability of neural networks has been extensively investigated. However, most existing results focus on the fault-free neural networks only. This papers extends the analysis on the generalization ability for networks with multiplicative weight noise. Our analysis result allows us not only to estimate the generalization ability of a faulty network, but also to select a good model from various settings. Simulated experiments are performed to verify theoretical result. {\textcopyright} 2009 Springer Berlin Heidelberg.",

author = "Leung, \{Chi Sing\} and Sum, \{Pui Fai\} and Hongjiang Wang",

year = "2009",

doi = "10.1007/978-3-642-03040-6\_39",

language = "English",

isbn = "3642030394",

volume = "5507 LNCS",

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

publisher = "Springer Verlag",

pages = "316--323",

booktitle = "Advances in Neuro-Information Processing",

address = "Germany",

edition = "PART 2",

note = "15th International Conference on Neuro-Information Processing, ICONIP 2008 ; Conference date: 25-11-2008 Through 28-11-2008",

}

Leung, CS, Sum, PF & Wang, H 2009, Analysis on generalization error of faulty rbf networks with weight decay regularizer. in Advances in Neuro-Information Processing: 15th International Conference, ICONIP 2008, Revised Selected Papers. PART 2 edn, vol. 5507 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 5507 LNCS, Springer Verlag, pp. 316-323, 15th International Conference on Neuro-Information Processing, ICONIP 2008, Auckland, New Zealand, 25/11/08. https://doi.org/10.1007/978-3-642-03040-6_39

Analysis on generalization error of faulty rbf networks with weight decay regularizer. / Leung, Chi Sing; Sum, Pui Fai; Wang, Hongjiang.
Advances in Neuro-Information Processing: 15th International Conference, ICONIP 2008, Revised Selected Papers. Vol. 5507 LNCS PART 2. ed. Springer Verlag, 2009. p. 316-323 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5507 LNCS).

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

TY - GEN

T1 - Analysis on generalization error of faulty rbf networks with weight decay regularizer

AU - Leung, Chi Sing

AU - Sum, Pui Fai

AU - Wang, Hongjiang

PY - 2009

Y1 - 2009

N2 - In the past two decades, the use of the weight decay regularizer for improving the generalization ability of neural networks has been extensively investigated. However, most existing results focus on the fault-free neural networks only. This papers extends the analysis on the generalization ability for networks with multiplicative weight noise. Our analysis result allows us not only to estimate the generalization ability of a faulty network, but also to select a good model from various settings. Simulated experiments are performed to verify theoretical result. © 2009 Springer Berlin Heidelberg.

AB - In the past two decades, the use of the weight decay regularizer for improving the generalization ability of neural networks has been extensively investigated. However, most existing results focus on the fault-free neural networks only. This papers extends the analysis on the generalization ability for networks with multiplicative weight noise. Our analysis result allows us not only to estimate the generalization ability of a faulty network, but also to select a good model from various settings. Simulated experiments are performed to verify theoretical result. © 2009 Springer Berlin Heidelberg.

UR - https://www.scopus.com/pages/publications/70349117707

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-70349117707&origin=recordpage

U2 - 10.1007/978-3-642-03040-6_39

DO - 10.1007/978-3-642-03040-6_39

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

SN - 3642030394

SN - 9783642030390

VL - 5507 LNCS

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

SP - 316

EP - 323

BT - Advances in Neuro-Information Processing

PB - Springer Verlag

T2 - 15th International Conference on Neuro-Information Processing, ICONIP 2008

Y2 - 25 November 2008 through 28 November 2008

ER -

Leung CS, Sum PF, Wang H. Analysis on generalization error of faulty rbf networks with weight decay regularizer. In Advances in Neuro-Information Processing: 15th International Conference, ICONIP 2008, Revised Selected Papers. PART 2 ed. Vol. 5507 LNCS. Springer Verlag. 2009. p. 316-323. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-642-03040-6_39

Analysis on generalization error of faulty rbf networks with weight decay regularizer

Abstract

Publication series

Conference

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this