Abstract
The non-stationary nature of electroencephalography (EEG) signals makes an EEG-based brain-computer interface (BCI) a dynamic system, thus improving its performance is a challenging task. In addition, it is well-known that due to non-stationarity based covariate shifts, the input data distributions of EEG-based BCI systems change during inter- and intra-session transitions, which poses great difficulty for developments of online adaptive data-driven systems. Ensemble learning approaches have been used previously to tackle this challenge. However, passive scheme based implementation leads to poor efficiency while increasing high computational cost. This paper presents a novel integration of covariate shift estimation and unsupervised adaptive ensemble learning (CSE-UAEL) to tackle non-stationarity in motor-imagery (MI) related EEG classification. The proposed method first employs an exponentially weighted moving average model to detect the covariate shifts in the common spatial pattern features extracted from MI related brain responses. Then, a classifier ensemble was created and updated over time to account for changes in streaming input data distribution wherein new classifiers are added to the ensemble in accordance with estimated shifts. Furthermore, using two publicly available BCI-related EEG datasets, the proposed method was extensively compared with the state-of-the-art single-classifier based passive scheme, single-classifier based active scheme and ensemble based passive schemes. The experimental results show that the proposed active scheme based ensemble learning algorithm significantly enhances the BCI performance in MI classifications.
Highlights
Acronyms: brain-computer interface (BCI), Brain-computer-interface; covariate shifts (CSs), Covariate shift; CSP, Common spatial pattern; CSA, Covariate shift adaptation; CS estimation (CSE), Covariate shift estimation; CSEUAEL, CSE-based unsupervised adaptive ensemble learning; CSV, Covariate shift validation; CSW, Covariate shift warning; dynamically weighted ensemble classification (DWEC), Dynamically weighted ensemble classification; EEG, Electroencephalography; ERD, Synchronization; ERS, Desynchronization; frequency bands (FBs), Frequency band; filter bank CSP (FBCSP), Filter bank common spatial pattern; EWMA, exponential weighted moving average; KNN, K-nearest-neighbors; linear discriminant analysis (LDA), Linear discriminant analysis; MI, Motor imagery; nonstationary learning (NSL), Non-stationary learning; principal component analysis (PCA), Principal component analysis; probabilistic weighted K nearest neighbour (PWKNN), Probabilistic weighted K-nearest neighbour; Random Subspace Method (RSM), Random subspace method; SSL, Semi-supervised learning.H
The aim of this paper is to extend our previous work and present a novel active scheme based unsupervised adaptive ensemble learning algorithm to adapt to CSs under non-stationary environments in EEG-based BCI systems
We have shown that a single active inductive classifier in single-trial EEG classification outperformed the existing passive scheme, the developed system was only applicable for the rehabilitative BCI systems
Summary
Acronyms: BCI, Brain-computer-interface; CS, Covariate shift; CSP, Common spatial pattern; CSA, Covariate shift adaptation; CSE, Covariate shift estimation; CSEUAEL, CSE-based unsupervised adaptive ensemble learning; CSV, Covariate shift validation; CSW, Covariate shift warning; DWEC, Dynamically weighted ensemble classification; EEG, Electroencephalography; ERD, Synchronization; ERS, Desynchronization; FB, Frequency band; FBCSP, Filter bank common spatial pattern; EWMA, exponential weighted moving average; KNN, K-nearest-neighbors; LDA, Linear discriminant analysis; MI, Motor imagery; NSL, Non-stationary learning; PCA, Principal component analysis; PWKNN, Probabilistic weighted K-nearest neighbour; RSM, Random subspace method; SSL, Semi-supervised learning.H. There exits a range of literature on transfer learning and domain adaptation theory, which aims to adapt to NSEs by transferring knowledge between training and test domains. In this case, one can match the features distribution of training and testing by the density ratio estimation approaches such as kernel mean matching [5], Kullback–Leibler importance estimation procedure, and leastsquares importance fitting [6]. To favorably transfer knowledge between domains, one needs to estimate the primary causal mechanism of the data generating process These methods have, a limited applicability in real world problems, where the data in test domain are generated while operating in real-time.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
