Toward Unsupervised Adaptation of LDA for Brain–Computer Interfaces

Vidaurre, Carmen; Kawanabe, Motoaki; Bünau, Paul von; Blankertz, Benjamin; Müller, Klaus‐Robert

doi:10.1109/tbme.2010.2093133

Cited by 267 publications

(276 citation statements)

References 24 publications

Supporting

Mentioning

272

Contrasting

Unclassified

Order By: Relevance

“…In this context, several asynchronous approaches have resulted in supervised and non-supervised learning techniques to cope with the changes in the distributions due to sessionand user-related non-stationarities [5], [7], [8]. In a similar way, past studies with synchronous signals have found userrelated differences in the averaged P300 ERP component from the neurophysiological point of view [6].…”

Section: Introductionmentioning

confidence: 93%

Minimizing calibration time using inter-subject information of single-trial recognition of error potentials in brain-computer interfaces

Iturrate

Montesano

Chavarriaga

et al. 2011

2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society

View full text Add to dashboard Cite

Abstract-One of the main problems of both synchronous and asynchronous EEG-based BCIs is the need of an initial calibration phase before the system can be used. This phase is necessary due to the high non-stationarity of the EEG, since it changes between sessions and users. The calibration process limits the BCI systems to scenarios where the outputs are very controlled, and makes these systems non-friendly and exhausting for the users. Although it has been studied how to reduce calibration time for asynchronous signals, it is still an open issue for event-related potentials. Here, we propose the minimization of the calibration time on singletrial error potentials by using classifiers based on inter-subject information. The results show that it is possible to have a classifier with a high performance from the beginning of the experiment, and which is able to adapt itself making the calibration phase shorter and transparent to the user.

show abstract

Section: Introductionmentioning

confidence: 93%

Minimizing calibration time using inter-subject information of single-trial recognition of error potentials in brain-computer interfaces

Iturrate

Montesano

Chavarriaga

et al. 2011

2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society

View full text Add to dashboard Cite

show abstract

“…The bias b = w Tμ can be calculated without class information based on the estimated pooled meanμ = (μ 1 +μ 2 ) /2, if both classes can assumed to the equally sized (see [9]). …”

Section: Ldamentioning

confidence: 99%

“…The bias of the LDA in (2) was adapted by updating the pooled meanμ after each trial. For more information on LDA unsupervised adaptation see [9].…”

Section: F Runs 5: Unsupervisedmentioning

confidence: 99%

“…Brain Computer Interfaces (BCI) based on sensorimotor rhythms (SMRs), make use of the voluntary modulation in the alpha (8)(9)(10)(11)(12) or beta (13-20 Hz) frequency band of the electroencephalography (EEG) activity over sensorimotor areas during limb movement imagination to obtain, by proper real time processing of the brain activity, a control signal for an external device. Despite a great progress in BCI research (see [1] for a review), still 20-30% of the healthy population is not able to reach the level criterion of 70% of accuracy, above which users feel to have obtained BCI control, as assessed in [2] by a psychological study for two-class BCI with communication applications.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Common Spatial Pattern Patches: Online evaluation on BCI-naive users

Sannelli

Vidaurre

Müller

et al. 2012

2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society

Self Cite

View full text Add to dashboard Cite

Abstract-Brain-Computer Interfaces (BCI) based on the voluntary modulation of sensorimotor rhythms (SMRs) induced by motor imagery are very prominent because allow a continuous control of the external device. Nevertheless, the design of a SMR-based BCI system that provides every user with a reliable BCI control from the first session, i.e., without extensive training, is still a big challenge. Considerable advances in this direction have been made by the machine learning coadaptive calibration approach, which combines online adaptation techniques with subject learning in order to offer the user a feedback from the beginning of the experiment. Recently, based on offline analyses, we proposed the novel Common Spatial Patterns Patches (CSPP) technique as a good candidate to improve the co-adaptive calibration. CSPP is an ensemble of localized spatial filters, each of them optimized on subjectspecific data by CSP analysis. Here, the evaluation of CSPP in online operation is presented for the first time. Results on three BCI-naive participants show indeed promising results. All three users reach the threshold criterion of 70% accuracy within one session, even one candidate for whom the weak SMR at rest predicted deficient BCI control. Concurrent recordings of the SMR during a relax condition as well as the course of BCI performance indicate a clear learning effect.

show abstract

“…This problem arises of a variety of applications, such as computer vision [6][7][8][9], multivariate time series [10], and sentiment analysis [11,12]. In this paper, an invariant latent space is extracted to tackle domain shift problem.…”

Section: Introductionmentioning

confidence: 99%

DiReT: An effective discriminative dimensionality reduction approach for multi-source transfer learning

Tahmoresnezhad

Hashemi

2017

Scientia Iranica

View full text Add to dashboard Cite

KEYWORDSMulti-source transfer learning; Domain adaptation; Discriminative dimensionality reduction; Fisher discriminant analysis.Abstract. Transfer learning is a well-known solution to the problem of domain shift in which source domain (training set) and target domain (test set) are drawn from di erent distributions. In the absence of domain shift, discriminative dimensionality reduction approaches could classify target data with acceptable accuracy. However, distribution di erence across source and target domains degrades the performance of dimensionality reduction methods. In this paper, we propose a Discriminative Dimensionality Reduction approach for multi-source Transfer learning, DiReT, in which discrimination is exploited on transferred data. DiReT nds an embedded space, such that the distribution di erence of the source and target domains is minimized. Moreover, DiReT employs multiple source domains and semi-supervised target domain to transfer knowledge from multiple resources, and it also bridges across source and target domains to nd common knowledge in an embedded space. Empirical evidence of real and arti cial datasets indicates that DiReT manages to improve substantially over dimensionality reduction approaches.

show abstract

Toward Unsupervised Adaptation of LDA for Brain–Computer Interfaces

Cited by 267 publications

References 24 publications

Minimizing calibration time using inter-subject information of single-trial recognition of error potentials in brain-computer interfaces

Minimizing calibration time using inter-subject information of single-trial recognition of error potentials in brain-computer interfaces

Common Spatial Pattern Patches: Online evaluation on BCI-naive users

DiReT: An effective discriminative dimensionality reduction approach for multi-source transfer learning

Contact Info

Product

Resources

About