Detection of transient EEG patterns with adaptive unsupervised neural networks

Ozcan Ozdamar, Carlos N. Lopez, Ilker Yaylali

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

In this study, adaptive resonance theory (ART2) neural networks are investigated for on-line unsupervised recognition of EEG.spikes for epilepsy monitoring. ART2 networks are unsupervised self-organizing systems that cluster data into different classes. Learning is completed when these classes are labelled bv an expert and are saved in a look-up table for use by the system. Recognition of new data is accomplished by finding the class assigned by ART2 and searching through the table to get the proper labels. Unrecognized inputs are put into a new class for future labelling. In this study an ART2 neural network with 20 inputs was developed and trained using EEG data containing spike and non-spike waveforms. For comparison a 20 input multilayer perceptron was constructed and evaluated similarly. Evaluation with three sets of EEG patterns indicates that the ART2 network's performance is close to that of multilayer perceptron showing its high potential. Considering that ART2 can be trained with one or a few iterations (compared to thousands required for hackpropagation networks), monitoring systems with on-line training can be easily constructed. Such systems are highly desirable for long-term EEG monitoring due to large variations of spike waveforms that occur among individual patients and during long recording time periods.

Original languageEnglish (US)
Title of host publicationProceedings of the 1992 International Biomedical Engineering Days, IBED 1992
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages192-197
Number of pages6
ISBN (Electronic)0780307437, 9780780307438
DOIs
StatePublished - Jan 1 1992
Event1992 International Biomedical Engineering Days, IBED 1992 - Istanbul, Turkey
Duration: Aug 18 1992Aug 20 1992

Publication series

NameProceedings of the 1992 International Biomedical Engineering Days, IBED 1992

Conference

Conference1992 International Biomedical Engineering Days, IBED 1992
CountryTurkey
CityIstanbul
Period8/18/928/20/92

Fingerprint

Electroencephalography
Neural networks
Multilayer neural networks
Monitoring
Network performance
Labeling
Labels

ASJC Scopus subject areas

  • Biomedical Engineering

Cite this

Ozdamar, O., Lopez, C. N., & Yaylali, I. (1992). Detection of transient EEG patterns with adaptive unsupervised neural networks. In Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992 (pp. 192-197). [247111] (Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IBED.1992.247111

Detection of transient EEG patterns with adaptive unsupervised neural networks. / Ozdamar, Ozcan; Lopez, Carlos N.; Yaylali, Ilker.

Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992. Institute of Electrical and Electronics Engineers Inc., 1992. p. 192-197 247111 (Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Ozdamar, O, Lopez, CN & Yaylali, I 1992, Detection of transient EEG patterns with adaptive unsupervised neural networks. in Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992., 247111, Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992, Institute of Electrical and Electronics Engineers Inc., pp. 192-197, 1992 International Biomedical Engineering Days, IBED 1992, Istanbul, Turkey, 8/18/92. https://doi.org/10.1109/IBED.1992.247111
Ozdamar O, Lopez CN, Yaylali I. Detection of transient EEG patterns with adaptive unsupervised neural networks. In Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992. Institute of Electrical and Electronics Engineers Inc. 1992. p. 192-197. 247111. (Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992). https://doi.org/10.1109/IBED.1992.247111
Ozdamar, Ozcan ; Lopez, Carlos N. ; Yaylali, Ilker. / Detection of transient EEG patterns with adaptive unsupervised neural networks. Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992. Institute of Electrical and Electronics Engineers Inc., 1992. pp. 192-197 (Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992).
@inproceedings{e3216ab4fa7140ca9d90ddf7ad438a78,
title = "Detection of transient EEG patterns with adaptive unsupervised neural networks",
abstract = "In this study, adaptive resonance theory (ART2) neural networks are investigated for on-line unsupervised recognition of EEG.spikes for epilepsy monitoring. ART2 networks are unsupervised self-organizing systems that cluster data into different classes. Learning is completed when these classes are labelled bv an expert and are saved in a look-up table for use by the system. Recognition of new data is accomplished by finding the class assigned by ART2 and searching through the table to get the proper labels. Unrecognized inputs are put into a new class for future labelling. In this study an ART2 neural network with 20 inputs was developed and trained using EEG data containing spike and non-spike waveforms. For comparison a 20 input multilayer perceptron was constructed and evaluated similarly. Evaluation with three sets of EEG patterns indicates that the ART2 network's performance is close to that of multilayer perceptron showing its high potential. Considering that ART2 can be trained with one or a few iterations (compared to thousands required for hackpropagation networks), monitoring systems with on-line training can be easily constructed. Such systems are highly desirable for long-term EEG monitoring due to large variations of spike waveforms that occur among individual patients and during long recording time periods.",
author = "Ozcan Ozdamar and Lopez, {Carlos N.} and Ilker Yaylali",
year = "1992",
month = "1",
day = "1",
doi = "10.1109/IBED.1992.247111",
language = "English (US)",
series = "Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "192--197",
booktitle = "Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992",

}

TY - GEN

T1 - Detection of transient EEG patterns with adaptive unsupervised neural networks

AU - Ozdamar, Ozcan

AU - Lopez, Carlos N.

AU - Yaylali, Ilker

PY - 1992/1/1

Y1 - 1992/1/1

N2 - In this study, adaptive resonance theory (ART2) neural networks are investigated for on-line unsupervised recognition of EEG.spikes for epilepsy monitoring. ART2 networks are unsupervised self-organizing systems that cluster data into different classes. Learning is completed when these classes are labelled bv an expert and are saved in a look-up table for use by the system. Recognition of new data is accomplished by finding the class assigned by ART2 and searching through the table to get the proper labels. Unrecognized inputs are put into a new class for future labelling. In this study an ART2 neural network with 20 inputs was developed and trained using EEG data containing spike and non-spike waveforms. For comparison a 20 input multilayer perceptron was constructed and evaluated similarly. Evaluation with three sets of EEG patterns indicates that the ART2 network's performance is close to that of multilayer perceptron showing its high potential. Considering that ART2 can be trained with one or a few iterations (compared to thousands required for hackpropagation networks), monitoring systems with on-line training can be easily constructed. Such systems are highly desirable for long-term EEG monitoring due to large variations of spike waveforms that occur among individual patients and during long recording time periods.

AB - In this study, adaptive resonance theory (ART2) neural networks are investigated for on-line unsupervised recognition of EEG.spikes for epilepsy monitoring. ART2 networks are unsupervised self-organizing systems that cluster data into different classes. Learning is completed when these classes are labelled bv an expert and are saved in a look-up table for use by the system. Recognition of new data is accomplished by finding the class assigned by ART2 and searching through the table to get the proper labels. Unrecognized inputs are put into a new class for future labelling. In this study an ART2 neural network with 20 inputs was developed and trained using EEG data containing spike and non-spike waveforms. For comparison a 20 input multilayer perceptron was constructed and evaluated similarly. Evaluation with three sets of EEG patterns indicates that the ART2 network's performance is close to that of multilayer perceptron showing its high potential. Considering that ART2 can be trained with one or a few iterations (compared to thousands required for hackpropagation networks), monitoring systems with on-line training can be easily constructed. Such systems are highly desirable for long-term EEG monitoring due to large variations of spike waveforms that occur among individual patients and during long recording time periods.

UR - http://www.scopus.com/inward/record.url?scp=0003126669&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0003126669&partnerID=8YFLogxK

U2 - 10.1109/IBED.1992.247111

DO - 10.1109/IBED.1992.247111

M3 - Conference contribution

T3 - Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992

SP - 192

EP - 197

BT - Proceedings of the 1992 International Biomedical Engineering Days, IBED 1992

PB - Institute of Electrical and Electronics Engineers Inc.

ER -