Ítem
Solo Metadatos

Incremental Training of Neural Network for Motor Tasks Recognition Based on Brain-Computer Interface

Mostrar el registro sencillo de la publicación
dc.creatorTriana Guzmán N.spa
dc.creatorOrjuela Cañón, Alvaro Davidspa
dc.creatorJutinico Alarcon A.L.spa
dc.date.accessioned2020-05-25T23:56:49Z
dc.date.available2020-05-25T23:56:49Z
dc.date.created2019spa
dc.description.abstractBrain-computer interfaces (BCI) based on motor imagery tasks (MI) have been established as a promising solution for restoring communication and control of people with motor disabilities. Physically impaired people may perform different motor imagery tasks which could be recorded in a non-invasive way using electroencephalography (EEG). However, the success of the MI-BCI systems depends on the reliable processing of the EEG signals and the adequate selection of the features used to characterize the brain activity signals for effective classification of MI activity and translation into corresponding actions. The multilayer perceptron (MLP) has been the neural network most widely used for classification in BCI technologies. The fact that MLP is a universal approximator makes this classifier sensitive to overtraining, especially with such noisy, non-linear, and non-stationary data as EEG. Traditional training techniques, as well as more recent ones, have mainly focused on the machine-learning aspects of BCI training. As a novel alternative for BCI training, this work proposes an incremental training process. Preliminary results with a non-disabled individual demonstrate that the proposed method has been able to improve the BCI training performance in comparison with the cross-validation technique. Best results showed that the incremental training proposal allowed an increase of the performance by at least 10% in terms of classification compared to a conventional cross-validation technique, which indicates the potential application for classification models of BCI’s systems. © Springer Nature Switzerland AG 2019.eng
dc.format.mimetypeapplication/pdf
dc.identifier.doihttps://doi.org/10.1007/978-3-030-33904-3_57
dc.identifier.urihttps://repository.urosario.edu.co/handle/10336/22530
dc.language.isoengspa
dc.publisherSpringerspa
dc.relation.citationEndPage619
dc.relation.citationStartPage610
dc.relation.citationTitleLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
dc.relation.citationVolumeVol. 11896 LNCS
dc.relation.ispartofLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol.11896 LNCS,(2019); pp. 610-619spa
dc.relation.urihttps://www.scopus.com/inward/record.uri?eid=2-s2.0-85075643414&doi=10.1007%2f978-3-030-33904-3_57&partnerID=40&md5=6e8f3917b8f9271a0395052a588f57ecspa
dc.rights.accesRightsinfo:eu-repo/semantics/openAccess
dc.rights.accesoAbierto (Texto Completo)spa
dc.source.instnameinstname:Universidad del Rosariospa
dc.source.reponamereponame:Repositorio Institucional EdocURspa
dc.subject.keywordBiomedical signal processingspa
dc.subject.keywordBrainspa
dc.subject.keywordElectroencephalographyspa
dc.subject.keywordElectrophysiologyspa
dc.subject.keywordMultilayer neural networksspa
dc.subject.keywordMultilayersspa
dc.subject.keywordPattern recognitionspa
dc.subject.keywordClassification modelsspa
dc.subject.keywordCommunication and controlspa
dc.subject.keywordCross validationspa
dc.subject.keywordCross-validation techniquespa
dc.subject.keywordIncremental trainingspa
dc.subject.keywordMotor imageryspa
dc.subject.keywordMulti layer perceptronspa
dc.subject.keywordUniversal approximatorsspa
dc.subject.keywordBrain computer interfacespa
dc.subject.keywordBrain-computer interfacespa
dc.subject.keywordCross-validationspa
dc.subject.keywordElectroencephalographyspa
dc.subject.keywordIncremental trainingspa
dc.subject.keywordMotor imageryspa
dc.subject.keywordMultilayer perceptronspa
dc.titleIncremental Training of Neural Network for Motor Tasks Recognition Based on Brain-Computer Interfacespa
dc.typeconferenceObjecteng
dc.type.hasVersioninfo:eu-repo/semantics/publishedVersion
dc.type.spaDocumento de conferenciaspa
Archivos
Colecciones
Artículos