An incremental learning algorithm with confidence estimation for automated identification of NDE signals

Robi Polikar, Lalita Udpa, Satish Udpa, Vasant Honavar

Research output: Contribution to journalArticle

27 Citations (Scopus)

Abstract

An incremental learning algorithm is introduced for learning new information from additional data that may later become available, after a classifier has already been trained using a previously available database. The proposed algorithm is capable of incrementally learning new information without forgetting previously acquired knowledge and without requiring access to the original database, even when new data include examples of previously unseen classes. Scenarios requiring such a learning algorithm are encountered often in nondestructive evaluation (NDE) in which large volumes of data are collected in batches over a period of time, and new defect types may become available in subsequent databases. The algorithm, named Learn++, takes advantage of synergistic generalization performance of an ensemble of classifiers in which each classifier is trained with a strategically chosen subset of the training databases that subsequently become available. The ensemble of classifiers then is combined through a weighted majority voting procedure. Learn++ is independent of the specific classifier(s) comprising the ensemble, and hence may be used with any supervised learning algorithm. The voting procedure also allows Learn++ to estimate the confidence in its own decision. We present the algorithm and its promising results on two separate ultrasonic weld inspection applications.

Original languageEnglish (US)
Pages (from-to)990-1001
Number of pages12
JournalIEEE transactions on ultrasonics, ferroelectrics, and frequency control
Volume51
Issue number8
DOIs
StatePublished - Aug 1 2004

Fingerprint

Learning algorithms
learning
confidence
classifiers
Classifiers
evaluation
voting
Supervised learning
Welds
Inspection
Ultrasonics
set theory
inspection
education
ultrasonics
Defects
defects
estimates

All Science Journal Classification (ASJC) codes

  • Instrumentation
  • Acoustics and Ultrasonics
  • Electrical and Electronic Engineering

Cite this

@article{eeb1577207aa416aa5b5505db702db2f,
title = "An incremental learning algorithm with confidence estimation for automated identification of NDE signals",
abstract = "An incremental learning algorithm is introduced for learning new information from additional data that may later become available, after a classifier has already been trained using a previously available database. The proposed algorithm is capable of incrementally learning new information without forgetting previously acquired knowledge and without requiring access to the original database, even when new data include examples of previously unseen classes. Scenarios requiring such a learning algorithm are encountered often in nondestructive evaluation (NDE) in which large volumes of data are collected in batches over a period of time, and new defect types may become available in subsequent databases. The algorithm, named Learn++, takes advantage of synergistic generalization performance of an ensemble of classifiers in which each classifier is trained with a strategically chosen subset of the training databases that subsequently become available. The ensemble of classifiers then is combined through a weighted majority voting procedure. Learn++ is independent of the specific classifier(s) comprising the ensemble, and hence may be used with any supervised learning algorithm. The voting procedure also allows Learn++ to estimate the confidence in its own decision. We present the algorithm and its promising results on two separate ultrasonic weld inspection applications.",
author = "Robi Polikar and Lalita Udpa and Satish Udpa and Vasant Honavar",
year = "2004",
month = "8",
day = "1",
doi = "10.1109/TUFFC.2004.1324403",
language = "English (US)",
volume = "51",
pages = "990--1001",
journal = "IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control",
issn = "0885-3010",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "8",

}

An incremental learning algorithm with confidence estimation for automated identification of NDE signals. / Polikar, Robi; Udpa, Lalita; Udpa, Satish; Honavar, Vasant.

In: IEEE transactions on ultrasonics, ferroelectrics, and frequency control, Vol. 51, No. 8, 01.08.2004, p. 990-1001.

Research output: Contribution to journalArticle

TY - JOUR

T1 - An incremental learning algorithm with confidence estimation for automated identification of NDE signals

AU - Polikar, Robi

AU - Udpa, Lalita

AU - Udpa, Satish

AU - Honavar, Vasant

PY - 2004/8/1

Y1 - 2004/8/1

N2 - An incremental learning algorithm is introduced for learning new information from additional data that may later become available, after a classifier has already been trained using a previously available database. The proposed algorithm is capable of incrementally learning new information without forgetting previously acquired knowledge and without requiring access to the original database, even when new data include examples of previously unseen classes. Scenarios requiring such a learning algorithm are encountered often in nondestructive evaluation (NDE) in which large volumes of data are collected in batches over a period of time, and new defect types may become available in subsequent databases. The algorithm, named Learn++, takes advantage of synergistic generalization performance of an ensemble of classifiers in which each classifier is trained with a strategically chosen subset of the training databases that subsequently become available. The ensemble of classifiers then is combined through a weighted majority voting procedure. Learn++ is independent of the specific classifier(s) comprising the ensemble, and hence may be used with any supervised learning algorithm. The voting procedure also allows Learn++ to estimate the confidence in its own decision. We present the algorithm and its promising results on two separate ultrasonic weld inspection applications.

AB - An incremental learning algorithm is introduced for learning new information from additional data that may later become available, after a classifier has already been trained using a previously available database. The proposed algorithm is capable of incrementally learning new information without forgetting previously acquired knowledge and without requiring access to the original database, even when new data include examples of previously unseen classes. Scenarios requiring such a learning algorithm are encountered often in nondestructive evaluation (NDE) in which large volumes of data are collected in batches over a period of time, and new defect types may become available in subsequent databases. The algorithm, named Learn++, takes advantage of synergistic generalization performance of an ensemble of classifiers in which each classifier is trained with a strategically chosen subset of the training databases that subsequently become available. The ensemble of classifiers then is combined through a weighted majority voting procedure. Learn++ is independent of the specific classifier(s) comprising the ensemble, and hence may be used with any supervised learning algorithm. The voting procedure also allows Learn++ to estimate the confidence in its own decision. We present the algorithm and its promising results on two separate ultrasonic weld inspection applications.

UR - http://www.scopus.com/inward/record.url?scp=4043139306&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=4043139306&partnerID=8YFLogxK

U2 - 10.1109/TUFFC.2004.1324403

DO - 10.1109/TUFFC.2004.1324403

M3 - Article

C2 - 15344404

AN - SCOPUS:4043139306

VL - 51

SP - 990

EP - 1001

JO - IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control

JF - IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control

SN - 0885-3010

IS - 8

ER -