Adaptive KF-SVM Classification for Single Trial EEG in BCI

Single trial electroencephalogram classification is indispensable in online brain–computer interfaces (BCIs) A classification method called adaptive Kernel Fisher Support Vector Machine (KF-SVM) is designed and applied to single trial EEG classification in BCIs. The adaptive KF-SVM algorithm combines adaptive idea, SVM and within-class scatter inspired from kernel fisher. Firstly, the within-class scatter matrix of a feature vector is calculated. And to construct a new kernel, this scatter is incorporated into the kernel function of SVM. Ultimately, the recognition result is calculated by the SVM whose kernel has been changed. The proposed algorithm simultaneously maximizes the discrimination between classes and also considers the within-class dissimilarities, which avoids some disadvantages of traditional SVM. In addition, the within-class scatter matrix of adaptive KF-SVM is updated trial by trail, which enhances the online adaptation of BCIs. Based on the EEG data recorded from seven subjects, the new approach achieved higher classification accuracies than the standard SVM, KF-SVM and adaptive linear classifier. The proposed scheme achieves the average performance improvement of 5.8%,5.2% and 3.7% respectively compared to other three schemes.

[1]  Liborio Cavaleri,et al.  Prediction of the Fundamental Period of Infilled RC Frame Structures Using Artificial Neural Networks , 2015, Comput. Intell. Neurosci..

[2]  Dongrui Wu,et al.  Online and Offline Domain Adaptation for Reducing BCI Calibration Effort , 2017, IEEE Transactions on Human-Machine Systems.

[3]  Dingguo Zhang,et al.  Unsupervised adaptation of electroencephalogram signal processing based on fuzzy C-means algorithm , 2012 .

[4]  Wei-Yen Hsu,et al.  EEG-based motor imagery classification using enhanced active segment selection and adaptive classifier , 2011, Comput. Biol. Medicine.

[5]  Cuntai Guan,et al.  On the use of convolutional neural networks and augmented CSP features for multi-class motor imagery of EEG signals classification , 2015, 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC).

[6]  Tzyy-Ping Jung,et al.  An Online Brain-Computer Interface Based on SSVEPs Measured From Non-Hair-Bearing Areas , 2017, IEEE Transactions on Neural Systems and Rehabilitation Engineering.

[7]  Henrik Zetterberg,et al.  Neurofilament Light: A Dynamic Cross-Disease Fluid Biomarker for Neurodegeneration , 2016, Neuron.

[8]  Anna Lisa Mangia,et al.  EEG-Based BCI System Using Adaptive Features Extraction and Classification Procedures , 2016, Comput. Intell. Neurosci..

[9]  Yan Wu,et al.  A novel method for motor imagery EEG adaptive classification based biomimetic pattern recognition , 2013, Neurocomputing.

[10]  Wolfgang Rosenstiel,et al.  Adaptive SVM-Based Classification Increases Performance of a MEG-Based Brain-Computer Interface (BCI) , 2012, ICANN.

[11]  R. Bhavani,et al.  Automatic classification of computed tomography brain images using ANN, k-NN and SVM , 2013, AI & SOCIETY.

[12]  Cuntai Guan,et al.  Optimizing Spatial Filters by Minimizing Within-Class Dissimilarities in Electroencephalogram-Based Brain–Computer Interface , 2013, IEEE Transactions on Neural Networks and Learning Systems.

[13]  E Donchin,et al.  Brain-computer interface technology: a review of the first international meeting. , 2000, IEEE transactions on rehabilitation engineering : a publication of the IEEE Engineering in Medicine and Biology Society.

[14]  Elad Alon,et al.  Wireless Recording in the Peripheral Nervous System with Ultrasonic Neural Dust , 2016, Neuron.

[15]  Helge J. Ritter,et al.  BCI competition 2003-data set IIb: support vector machines for the P300 speller paradigm , 2004, IEEE Transactions on Biomedical Engineering.