Sleep stages classification from electroencephalographic signals based on unsupervised feature space clustering

Iosif Mporas, Anastasia Efstathiou, Vasileios Megalooikonomou

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

1 Citation (Scopus)

Abstract

In this article we present a methodology for the automatic classification of sleep stages. The methodology relies on short-time analysis with time and frequency domain features followed by unsupervised feature subspace clustering. For each cluster of the feature space a different classification setup is adopted thus fine-tuning the classification algorithm to the specifics of the corresponding feature subspace area. The experimental results showed that the proposed methodology achieved a sleep stage classification accuracy equal to 92.53%, which corresponds to an improvement of approximately 3% compared to the best performing single classifier without applying clustering of the feature space.

Original languageEnglish
Title of host publicationBrain Informatics and Health - 8th International Conference, BIH 2015, Proceedings
EditorsYike Guo Y., Sean Hill S., Karl Friston, Hanchuan Peng, Aldo Faisal A.
PublisherSpringer Nature
Pages77-85
Number of pages9
ISBN (Print)9783319233437
DOIs
Publication statusPublished - 1 Jan 2015
Externally publishedYes
Event8th International Conference on Brain Informatics and Health, BIH 2015 - London, United Kingdom
Duration: 30 Aug 20152 Sept 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9250
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference8th International Conference on Brain Informatics and Health, BIH 2015
Country/TerritoryUnited Kingdom
CityLondon
Period30/08/152/09/15

Keywords

  • Clustering
  • Electroencephalography
  • Sleep stages

Fingerprint

Dive into the research topics of 'Sleep stages classification from electroencephalographic signals based on unsupervised feature space clustering'. Together they form a unique fingerprint.

Cite this