EEG Sparse Representation Based Alertness States Identification Using Gini Index

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

*المؤلف المقابل لهذا العمل

نتاج البحث: Conference contribution

1 اقتباس (Scopus)

ملخص

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.

اللغة الأصليةEnglish
عنوان منشور المضيفNeural Information Processing - 25th International Conference, ICONIP 2018, Proceedings
المحررونSeiichi Ozawa, Andrew Chi Sing Leung, Long Cheng
ناشرSpringer Verlag
الصفحات478-488
عدد الصفحات11
رقم المعيار الدولي للكتب (المطبوع)9783030042387
المعرِّفات الرقمية للأشياء
حالة النشرPublished - 2018
الحدث25th International Conference on Neural Information Processing, ICONIP 2018 - Siem Reap, Cambodia
المدة: ديسمبر ١٣ ٢٠١٨ديسمبر ١٦ ٢٠١٨

سلسلة المنشورات

الاسمLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
مستوى الصوت11307 LNCS
رقم المعيار الدولي للدوريات (المطبوع)0302-9743
رقم المعيار الدولي للدوريات (الإلكتروني)1611-3349

Other

Other25th International Conference on Neural Information Processing, ICONIP 2018
الدولة/الإقليمCambodia
المدينةSiem Reap
المدة١٢/١٣/١٨١٢/١٦/١٨

ASJC Scopus subject areas

  • ???subjectarea.asjc.2600.2614???
  • ???subjectarea.asjc.1700.1700???

بصمة

أدرس بدقة موضوعات البحث “EEG Sparse Representation Based Alertness States Identification Using Gini Index'. فهما يشكلان معًا بصمة فريدة.

قم بذكر هذا