EEG Sparse Representation Based Alertness States Identification Using Gini Index

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

*Corresponding author for this work

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

1 Citation (Scopus)

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

Keywords

  • Alertness classification
  • Gini index
  • Sparse representation

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'EEG Sparse Representation Based Alertness States Identification Using Gini Index'. Together they form a unique fingerprint.

Cite this