EEG Sparse Representation Based Alertness States Identification Using Gini Index

Muna Tageldin, Talal Al-Mashaikki, Hamza Bali, Mostefa Mesbah

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

Abstract

Poor alertness experienced by individuals may lead to serious accidents that impact on people’s health and safety. To prevent such accidents, an efficient automatic alertness states identification is required. Sparse representation-based classification has recently gained a lot of popularity. A classifier from this class typically comprises three stages: dictionary learning, sparse coding and class assignment. Gini index, a recently proposed method, was shown to possess a number of properties that make it a better sparsity measure than the widely used l0- and l1-norms. This paper investigates whether these properties also lead to a better classifier. The proposed classifier, unlike the existing sparsity-based ones, embeds the Gini index in all stages of the classification process. To assess its performance, the new classifier was used to automatically identify three alertness levels, namely awake, drowsy, and sleep using EEG signal. The obtained results show that the new classifier outperforms those based on l0- and l1-norms.

Original languageEnglish
Title of host publicationNeural Information Processing - 25th International Conference, ICONIP 2018, Proceedings
EditorsSeiichi Ozawa, Andrew Chi Sing Leung, Long Cheng
PublisherSpringer-Verlag
Pages478-488
Number of pages11
ISBN (Print)9783030042387
DOIs
Publication statusPublished - Jan 1 2018
Event25th International Conference on Neural Information Processing, ICONIP 2018 - Siem Reap, Cambodia
Duration: Dec 13 2018Dec 16 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11307 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other25th International Conference on Neural Information Processing, ICONIP 2018
CountryCambodia
CitySiem Reap
Period12/13/1812/16/18

Fingerprint

Gini Index
Sparse Representation
Electroencephalography
Classifiers
Classifier
L1-norm
Sparsity
Accidents
Sparse Coding
Sleep
Glossaries
Health
Assignment
Safety
Electroencephalogram

Keywords

  • Alertness classification
  • Gini index
  • Sparse representation

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Tageldin, M., Al-Mashaikki, T., Bali, H., & Mesbah, M. (2018). EEG Sparse Representation Based Alertness States Identification Using Gini Index. In S. Ozawa, A. C. S. Leung, & L. Cheng (Eds.), Neural Information Processing - 25th International Conference, ICONIP 2018, Proceedings (pp. 478-488). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11307 LNCS). Springer-Verlag. https://doi.org/10.1007/978-3-030-04239-4_43

EEG Sparse Representation Based Alertness States Identification Using Gini Index. / Tageldin, Muna; Al-Mashaikki, Talal; Bali, Hamza; Mesbah, Mostefa.

Neural Information Processing - 25th International Conference, ICONIP 2018, Proceedings. ed. / Seiichi Ozawa; Andrew Chi Sing Leung; Long Cheng. Springer-Verlag, 2018. p. 478-488 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11307 LNCS).

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

Tageldin, M, Al-Mashaikki, T, Bali, H & Mesbah, M 2018, EEG Sparse Representation Based Alertness States Identification Using Gini Index. in S Ozawa, ACS Leung & L Cheng (eds), Neural Information Processing - 25th International Conference, ICONIP 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11307 LNCS, Springer-Verlag, pp. 478-488, 25th International Conference on Neural Information Processing, ICONIP 2018, Siem Reap, Cambodia, 12/13/18. https://doi.org/10.1007/978-3-030-04239-4_43
Tageldin M, Al-Mashaikki T, Bali H, Mesbah M. EEG Sparse Representation Based Alertness States Identification Using Gini Index. In Ozawa S, Leung ACS, Cheng L, editors, Neural Information Processing - 25th International Conference, ICONIP 2018, Proceedings. Springer-Verlag. 2018. p. 478-488. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-04239-4_43
Tageldin, Muna ; Al-Mashaikki, Talal ; Bali, Hamza ; Mesbah, Mostefa. / EEG Sparse Representation Based Alertness States Identification Using Gini Index. Neural Information Processing - 25th International Conference, ICONIP 2018, Proceedings. editor / Seiichi Ozawa ; Andrew Chi Sing Leung ; Long Cheng. Springer-Verlag, 2018. pp. 478-488 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{2e9c87ccec484a20a4328c7bc4c9cdd9,
title = "EEG Sparse Representation Based Alertness States Identification Using Gini Index",
abstract = "Poor alertness experienced by individuals may lead to serious accidents that impact on people’s health and safety. To prevent such accidents, an efficient automatic alertness states identification is required. Sparse representation-based classification has recently gained a lot of popularity. A classifier from this class typically comprises three stages: dictionary learning, sparse coding and class assignment. Gini index, a recently proposed method, was shown to possess a number of properties that make it a better sparsity measure than the widely used l0- and l1-norms. This paper investigates whether these properties also lead to a better classifier. The proposed classifier, unlike the existing sparsity-based ones, embeds the Gini index in all stages of the classification process. To assess its performance, the new classifier was used to automatically identify three alertness levels, namely awake, drowsy, and sleep using EEG signal. The obtained results show that the new classifier outperforms those based on l0- and l1-norms.",
keywords = "Alertness classification, Gini index, Sparse representation",
author = "Muna Tageldin and Talal Al-Mashaikki and Hamza Bali and Mostefa Mesbah",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/978-3-030-04239-4_43",
language = "English",
isbn = "9783030042387",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer-Verlag",
pages = "478--488",
editor = "Seiichi Ozawa and Leung, {Andrew Chi Sing} and Long Cheng",
booktitle = "Neural Information Processing - 25th International Conference, ICONIP 2018, Proceedings",

}

TY - GEN

T1 - EEG Sparse Representation Based Alertness States Identification Using Gini Index

AU - Tageldin, Muna

AU - Al-Mashaikki, Talal

AU - Bali, Hamza

AU - Mesbah, Mostefa

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Poor alertness experienced by individuals may lead to serious accidents that impact on people’s health and safety. To prevent such accidents, an efficient automatic alertness states identification is required. Sparse representation-based classification has recently gained a lot of popularity. A classifier from this class typically comprises three stages: dictionary learning, sparse coding and class assignment. Gini index, a recently proposed method, was shown to possess a number of properties that make it a better sparsity measure than the widely used l0- and l1-norms. This paper investigates whether these properties also lead to a better classifier. The proposed classifier, unlike the existing sparsity-based ones, embeds the Gini index in all stages of the classification process. To assess its performance, the new classifier was used to automatically identify three alertness levels, namely awake, drowsy, and sleep using EEG signal. The obtained results show that the new classifier outperforms those based on l0- and l1-norms.

AB - Poor alertness experienced by individuals may lead to serious accidents that impact on people’s health and safety. To prevent such accidents, an efficient automatic alertness states identification is required. Sparse representation-based classification has recently gained a lot of popularity. A classifier from this class typically comprises three stages: dictionary learning, sparse coding and class assignment. Gini index, a recently proposed method, was shown to possess a number of properties that make it a better sparsity measure than the widely used l0- and l1-norms. This paper investigates whether these properties also lead to a better classifier. The proposed classifier, unlike the existing sparsity-based ones, embeds the Gini index in all stages of the classification process. To assess its performance, the new classifier was used to automatically identify three alertness levels, namely awake, drowsy, and sleep using EEG signal. The obtained results show that the new classifier outperforms those based on l0- and l1-norms.

KW - Alertness classification

KW - Gini index

KW - Sparse representation

UR - http://www.scopus.com/inward/record.url?scp=85058989680&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85058989680&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-04239-4_43

DO - 10.1007/978-3-030-04239-4_43

M3 - Conference contribution

SN - 9783030042387

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

SP - 478

EP - 488

BT - Neural Information Processing - 25th International Conference, ICONIP 2018, Proceedings

A2 - Ozawa, Seiichi

A2 - Leung, Andrew Chi Sing

A2 - Cheng, Long

PB - Springer-Verlag

ER -