A novel information-gap technique to assess reliability of neural network-based damage detection

S.G. Pierce, K. Worden, G. Manson

Research output: Contribution to journalArticle

33 Citations (Scopus)

Abstract

The application of neural network classifiers to a damage detection problem is discussed within a framework of an interval arithmetic-based information-gap technique. Using this approach the robustness of trained classifiers to uncertainty in their input data was assessed. Conventional network training using a regularised Maximum Likelihood approach is discussed and compared with interval propagation applied as a tool to evaluate the robustness of a particular network. Concepts of interval-based worst-case error and opportunity are introduced to facilitate the analysis. The interval-based approach is further developed into a network selection procedure capable of significant improvements (up to 22%) in the worst-case error performance over a conventional network trained on crisp (single-valued) data.
LanguageEnglish
Pages96-111
Number of pages15
JournalJournal of Sound and Vibration
Volume293
Issue number1-2
DOIs
Publication statusPublished - 30 May 2006

Fingerprint

Damage detection
Classifiers
damage
Neural networks
intervals
Maximum likelihood
classifiers
education
propagation
Uncertainty

Keywords

  • uncertain system
  • defect detection
  • interval arithmetic
  • neural networks

Cite this

@article{79aa4dce3e004fd8b910d7c60e8ea4c5,
title = "A novel information-gap technique to assess reliability of neural network-based damage detection",
abstract = "The application of neural network classifiers to a damage detection problem is discussed within a framework of an interval arithmetic-based information-gap technique. Using this approach the robustness of trained classifiers to uncertainty in their input data was assessed. Conventional network training using a regularised Maximum Likelihood approach is discussed and compared with interval propagation applied as a tool to evaluate the robustness of a particular network. Concepts of interval-based worst-case error and opportunity are introduced to facilitate the analysis. The interval-based approach is further developed into a network selection procedure capable of significant improvements (up to 22{\%}) in the worst-case error performance over a conventional network trained on crisp (single-valued) data.",
keywords = "uncertain system, defect detection, interval arithmetic, neural networks",
author = "S.G. Pierce and K. Worden and G. Manson",
year = "2006",
month = "5",
day = "30",
doi = "10.1016/j.jsv.2005.09.029",
language = "English",
volume = "293",
pages = "96--111",
journal = "Journal of Sound and Vibration",
issn = "0022-460X",
number = "1-2",

}

A novel information-gap technique to assess reliability of neural network-based damage detection. / Pierce, S.G.; Worden, K.; Manson, G.

In: Journal of Sound and Vibration, Vol. 293, No. 1-2, 30.05.2006, p. 96-111.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A novel information-gap technique to assess reliability of neural network-based damage detection

AU - Pierce, S.G.

AU - Worden, K.

AU - Manson, G.

PY - 2006/5/30

Y1 - 2006/5/30

N2 - The application of neural network classifiers to a damage detection problem is discussed within a framework of an interval arithmetic-based information-gap technique. Using this approach the robustness of trained classifiers to uncertainty in their input data was assessed. Conventional network training using a regularised Maximum Likelihood approach is discussed and compared with interval propagation applied as a tool to evaluate the robustness of a particular network. Concepts of interval-based worst-case error and opportunity are introduced to facilitate the analysis. The interval-based approach is further developed into a network selection procedure capable of significant improvements (up to 22%) in the worst-case error performance over a conventional network trained on crisp (single-valued) data.

AB - The application of neural network classifiers to a damage detection problem is discussed within a framework of an interval arithmetic-based information-gap technique. Using this approach the robustness of trained classifiers to uncertainty in their input data was assessed. Conventional network training using a regularised Maximum Likelihood approach is discussed and compared with interval propagation applied as a tool to evaluate the robustness of a particular network. Concepts of interval-based worst-case error and opportunity are introduced to facilitate the analysis. The interval-based approach is further developed into a network selection procedure capable of significant improvements (up to 22%) in the worst-case error performance over a conventional network trained on crisp (single-valued) data.

KW - uncertain system

KW - defect detection

KW - interval arithmetic

KW - neural networks

U2 - 10.1016/j.jsv.2005.09.029

DO - 10.1016/j.jsv.2005.09.029

M3 - Article

VL - 293

SP - 96

EP - 111

JO - Journal of Sound and Vibration

T2 - Journal of Sound and Vibration

JF - Journal of Sound and Vibration

SN - 0022-460X

IS - 1-2

ER -