Limits...
Towards zero training for brain-computer interfacing.

Krauledat M, Tangermann M, Blankertz B, Müller KR - PLoS ONE (2008)

Bottom Line: The method takes advantage of knowledge collected in previous sessions: By a novel technique, prototypical spatial filters are determined which have better generalization properties compared to single-session filters.The feasibility of our novel approach is demonstrated with a series of online BCI experiments.Although performed without any calibration measurement at all, no loss of classification performance was observed.

View Article: PubMed Central - PubMed

Affiliation: Machine Learning Laboratory, Berlin Institute of Technology, Berlin, Germany. kraulem@cs.tu-berlin.de

ABSTRACT
Electroencephalogram (EEG) signals are highly subject-specific and vary considerably even between recording sessions of the same user within the same experimental paradigm. This challenges a stable operation of Brain-Computer Interface (BCI) systems. The classical approach is to train users by neurofeedback to produce fixed stereotypical patterns of brain activity. In the machine learning approach, a widely adapted method for dealing with those variances is to record a so called calibration measurement on the beginning of each session in order to optimize spatial filters and classifiers specifically for each subject and each day. This adaptation of the system to the individual brain signature of each user relieves from the need of extensive user training. In this paper we suggest a new method that overcomes the requirement of these time-consuming calibration recordings for long-term BCI users. The method takes advantage of knowledge collected in previous sessions: By a novel technique, prototypical spatial filters are determined which have better generalization properties compared to single-session filters. In particular, they can be used in follow-up sessions without the need to recalibrate the system. This way the calibration periods can be dramatically shortened or even completely omitted for these 'experienced' BCI users. The feasibility of our novel approach is demonstrated with a series of online BCI experiments. Although performed without any calibration measurement at all, no loss of classification performance was observed.

Show MeSH
Projection of CSP filters onto the (C−1)-dimensional hypersphere.Distances between filters are defined by the angles between the projected filters.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2500157&req=5

pone-0002967-g002: Projection of CSP filters onto the (C−1)-dimensional hypersphere.Distances between filters are defined by the angles between the projected filters.

Mentions: CSP filters are obtained as solutions of a generalized eigenvalue problem. Since every multiple of an eigenvector is again a solution to the eigenvalue problem every point in the space of CSP filters () on the line through a CSP filter point and the origin form an equivalence class (except for the origin itself). More precisely, it is sufficient to consider only normalized CSP vectors on the (C−1)-dimensional hypersphere (cf. figure 2).


Towards zero training for brain-computer interfacing.

Krauledat M, Tangermann M, Blankertz B, Müller KR - PLoS ONE (2008)

Projection of CSP filters onto the (C−1)-dimensional hypersphere.Distances between filters are defined by the angles between the projected filters.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2500157&req=5

pone-0002967-g002: Projection of CSP filters onto the (C−1)-dimensional hypersphere.Distances between filters are defined by the angles between the projected filters.
Mentions: CSP filters are obtained as solutions of a generalized eigenvalue problem. Since every multiple of an eigenvector is again a solution to the eigenvalue problem every point in the space of CSP filters () on the line through a CSP filter point and the origin form an equivalence class (except for the origin itself). More precisely, it is sufficient to consider only normalized CSP vectors on the (C−1)-dimensional hypersphere (cf. figure 2).

Bottom Line: The method takes advantage of knowledge collected in previous sessions: By a novel technique, prototypical spatial filters are determined which have better generalization properties compared to single-session filters.The feasibility of our novel approach is demonstrated with a series of online BCI experiments.Although performed without any calibration measurement at all, no loss of classification performance was observed.

View Article: PubMed Central - PubMed

Affiliation: Machine Learning Laboratory, Berlin Institute of Technology, Berlin, Germany. kraulem@cs.tu-berlin.de

ABSTRACT
Electroencephalogram (EEG) signals are highly subject-specific and vary considerably even between recording sessions of the same user within the same experimental paradigm. This challenges a stable operation of Brain-Computer Interface (BCI) systems. The classical approach is to train users by neurofeedback to produce fixed stereotypical patterns of brain activity. In the machine learning approach, a widely adapted method for dealing with those variances is to record a so called calibration measurement on the beginning of each session in order to optimize spatial filters and classifiers specifically for each subject and each day. This adaptation of the system to the individual brain signature of each user relieves from the need of extensive user training. In this paper we suggest a new method that overcomes the requirement of these time-consuming calibration recordings for long-term BCI users. The method takes advantage of knowledge collected in previous sessions: By a novel technique, prototypical spatial filters are determined which have better generalization properties compared to single-session filters. In particular, they can be used in follow-up sessions without the need to recalibrate the system. This way the calibration periods can be dramatically shortened or even completely omitted for these 'experienced' BCI users. The feasibility of our novel approach is demonstrated with a series of online BCI experiments. Although performed without any calibration measurement at all, no loss of classification performance was observed.

Show MeSH