mdh.sePublikationer
Ändra sökning
RefereraExporteraLänk till posten
Permanent länk

Direktlänk
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annat format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annat språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf
Empirical studies of multiobjective evolutionary algorithm in classifying neural oscillations to motor imagery
Mälardalens högskola, Akademin för innovation, design och teknik.
2019 (Engelska)Självständigt arbete på grundnivå (kandidatexamen), 10 poäng / 15 hpStudentuppsats (Examensarbete)
Abstract [en]

Brain-computer interfaces (BCIs) enables direct communication between a brain and a computer by recording and analyzing a subject’s neural activity in real-time. Research in BCI that classifies motor imagery (MI) activities are common in the literature due to its importance and applicability, e.g., stroke rehabilitation. Electroencephalography (EEG) is often used as the recording technique because its non-invasive, portable and have a relatively low cost. However, an EEG recording returns a vast number of features which must be reduced to decrease the computational time and complexity of the classifier. For this purpose, feature selection is often applied. In this study, a multiobjective evolutionary algorithm (MOEA) was used as feature selection in a high spatial and temporal feature set to (1) compare pairwise combinations of different objectives, (2) evaluate the relationship between the specific objective pair and their relation to model prediction accuracy, (3) compare multiobjective optimization versus a linear combination of the individual objectives. The results show that correlation feature selection (CFS) obtained the best performance between the evaluated objectives which were also more optimized than a linear combination of the individual objectives when classified with support vector machine (SVM).

Ort, förlag, år, upplaga, sidor
2019. , s. 53
Nationell ämneskategori
Datavetenskap (datalogi)
Identifikatorer
URN: urn:nbn:se:mdh:diva-44826OAI: oai:DiVA.org:mdh-44826DiVA, id: diva2:1336611
Ämne / kurs
Datavetenskap
Handledare
Examinatorer
Tillgänglig från: 2019-09-30 Skapad: 2019-07-09 Senast uppdaterad: 2019-09-30Bibliografiskt granskad

Open Access i DiVA

fulltext(4587 kB)30 nedladdningar
Filinformation
Filnamn FULLTEXT01.pdfFilstorlek 4587 kBChecksumma SHA-512
cde23dd96d11ba50a6cac634693cbd2203dc9e7cb1a51bae43cb8795af789cf38bce46494c6977e07e148053a7761916712079f8bb3b142b7a248d7df3775c22
Typ fulltextMimetyp application/pdf

Sök vidare i DiVA

Av författaren/redaktören
Parkkila, Christoffer
Av organisationen
Akademin för innovation, design och teknik
Datavetenskap (datalogi)

Sök vidare utanför DiVA

GoogleGoogle Scholar
Totalt: 30 nedladdningar
Antalet nedladdningar är summan av nedladdningar för alla fulltexter. Det kan inkludera t.ex tidigare versioner som nu inte längre är tillgängliga.

urn-nbn

Altmetricpoäng

urn-nbn
Totalt: 100 träffar
RefereraExporteraLänk till posten
Permanent länk

Direktlänk
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annat format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annat språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf