Number of the records: 1
Classification of Heterogeneous EEG Data by Combining Random Forests
- 1.
SYSNO ASEP 0318431 Document Type C - Proceedings Paper (int. conf.) R&D Document Type Conference Paper Title Classification of Heterogeneous EEG Data by Combining Random Forests Title Klasifikace heterogenních EEG dat kombinováním modelů získaných metodou Random Forests Author(s) Klaschka, Jan (UIVT-O) RID, SAI, ORCID Source Title Proceedings of IASC 2008. - Tokyo : Japanese Society of Computational Statistics, 2008 / Mizuta M. ; Nakano J. - ISBN 978-4-9904445-1-8 Pages s. 888-896 Number of pages 9 s. Publication form CD-ROM - CD-ROM Action IASC 2008 Event date 05.12.2008-08.12.2008 VEvent location Yokohama Country JP - Japan Event type WRD Language eng - English Country JP - Japan Keywords EEG classification ; somnolence ; random forests ; combining classifiers Subject RIV FH - Neurology R&D Projects 1F84B/042/520 GA MDS - Ministry of Transport (MD) ME 949 GA MŠMT - Ministry of Education, Youth and Sports (MEYS) CEZ AV0Z10300504 - UIVT-O (2005-2011) Annotation The focus of the paper is development of classification models capable of distinguishing, based on electroencephalography (EEG) data, somnolence (sleepiness) from other brain states typically met when driving a car. It is a part of a broader project aimed at prevention of damage caused by drivers' microsleeps. Random Forests (RF) method was chosen, on account of previous experience, as a base classification tool for the classification tasks studied. It is, however, not only routinely used: Various classification models tailored for a specific individual are constructed by combining a RF model derived from the individual's data with a model based on the data of the other suitably selected individuals. Several model combining strategies are described and results of their application on real-life EEG data in an extensive computational experiment are reported. Workplace Institute of Computer Science Contact Tereza Šírová, sirova@cs.cas.cz, Tel.: 266 053 800 Year of Publishing 2009
Number of the records: 1