Robustness of the Common Spatial Patterns algorithm in the BCI-pipeline

B. Reuderink, Mannes Poel

Abstract

When we want to use brain-computer interfaces (BCI) as an input modality for gaming, a short setup procedure is necessary. Therefore a user model has to be learned using small training sets. The common spatial patterns (CSP) algorithm is often used in BCI. In this work we investigate how the CSP algorithm generalizes when using small training sets, how the performance changes over time, and how well CSP generalizes over persons. Our results indicate that the CSP algorithm severely overfits on small training sets. The CSP algorithm often selects a small number of spatial filters that generalize poorly, which can have in impact on the classification performance. The generalization performance does not degrade over time, which is promising, but the signal does not seem to be stationary. In its current form, the CSP generalizes poorly over persons.
Original languageUndefined
Place of PublicationEnschede
PublisherCentre for Telematics and Information Technology (CTIT)
Number of pages5
StatePublished - Jul 2008

Publication series

NameCTIT Technical Report Series
PublisherCentre for Telematics and Information Technology, University of Twente
No.DTR08-9/TR-CTIT-08-52
ISSN (Print)1381-3625

Fingerprint

Brain computer interface

Keywords

  • METIS-251091
  • EWI-13085
  • IR-64884

Cite this

Reuderink, B., & Poel, M. (2008). Robustness of the Common Spatial Patterns algorithm in the BCI-pipeline. (CTIT Technical Report Series; No. DTR08-9/TR-CTIT-08-52). Enschede: Centre for Telematics and Information Technology (CTIT).

Reuderink, B.; Poel, Mannes / Robustness of the Common Spatial Patterns algorithm in the BCI-pipeline.

Enschede : Centre for Telematics and Information Technology (CTIT), 2008. 5 p. (CTIT Technical Report Series; No. DTR08-9/TR-CTIT-08-52).

Research output: ProfessionalReport

@book{45e4059439514fcf86a47d58228f5807,
title = "Robustness of the Common Spatial Patterns algorithm in the BCI-pipeline",
abstract = "When we want to use brain-computer interfaces (BCI) as an input modality for gaming, a short setup procedure is necessary. Therefore a user model has to be learned using small training sets. The common spatial patterns (CSP) algorithm is often used in BCI. In this work we investigate how the CSP algorithm generalizes when using small training sets, how the performance changes over time, and how well CSP generalizes over persons. Our results indicate that the CSP algorithm severely overfits on small training sets. The CSP algorithm often selects a small number of spatial filters that generalize poorly, which can have in impact on the classification performance. The generalization performance does not degrade over time, which is promising, but the signal does not seem to be stationary. In its current form, the CSP generalizes poorly over persons.",
keywords = "METIS-251091, EWI-13085, IR-64884",
author = "B. Reuderink and Mannes Poel",
year = "2008",
month = "7",
series = "CTIT Technical Report Series",
publisher = "Centre for Telematics and Information Technology (CTIT)",
number = "DTR08-9/TR-CTIT-08-52",
address = "Netherlands",

}

Reuderink, B & Poel, M 2008, Robustness of the Common Spatial Patterns algorithm in the BCI-pipeline. CTIT Technical Report Series, no. DTR08-9/TR-CTIT-08-52, Centre for Telematics and Information Technology (CTIT), Enschede.

Robustness of the Common Spatial Patterns algorithm in the BCI-pipeline. / Reuderink, B.; Poel, Mannes.

Enschede : Centre for Telematics and Information Technology (CTIT), 2008. 5 p. (CTIT Technical Report Series; No. DTR08-9/TR-CTIT-08-52).

Research output: ProfessionalReport

TY - BOOK

T1 - Robustness of the Common Spatial Patterns algorithm in the BCI-pipeline

AU - Reuderink,B.

AU - Poel,Mannes

PY - 2008/7

Y1 - 2008/7

N2 - When we want to use brain-computer interfaces (BCI) as an input modality for gaming, a short setup procedure is necessary. Therefore a user model has to be learned using small training sets. The common spatial patterns (CSP) algorithm is often used in BCI. In this work we investigate how the CSP algorithm generalizes when using small training sets, how the performance changes over time, and how well CSP generalizes over persons. Our results indicate that the CSP algorithm severely overfits on small training sets. The CSP algorithm often selects a small number of spatial filters that generalize poorly, which can have in impact on the classification performance. The generalization performance does not degrade over time, which is promising, but the signal does not seem to be stationary. In its current form, the CSP generalizes poorly over persons.

AB - When we want to use brain-computer interfaces (BCI) as an input modality for gaming, a short setup procedure is necessary. Therefore a user model has to be learned using small training sets. The common spatial patterns (CSP) algorithm is often used in BCI. In this work we investigate how the CSP algorithm generalizes when using small training sets, how the performance changes over time, and how well CSP generalizes over persons. Our results indicate that the CSP algorithm severely overfits on small training sets. The CSP algorithm often selects a small number of spatial filters that generalize poorly, which can have in impact on the classification performance. The generalization performance does not degrade over time, which is promising, but the signal does not seem to be stationary. In its current form, the CSP generalizes poorly over persons.

KW - METIS-251091

KW - EWI-13085

KW - IR-64884

M3 - Report

T3 - CTIT Technical Report Series

BT - Robustness of the Common Spatial Patterns algorithm in the BCI-pipeline

PB - Centre for Telematics and Information Technology (CTIT)

ER -

Reuderink B, Poel M. Robustness of the Common Spatial Patterns algorithm in the BCI-pipeline. Enschede: Centre for Telematics and Information Technology (CTIT), 2008. 5 p. (CTIT Technical Report Series; DTR08-9/TR-CTIT-08-52).