A Method for SSVEP Recognition Based on Weighted Canonical Correlation Analysis

As a multivariate statistical method, canonical correlation analysis (CCA) has been one of the most common methods for recognizing steady-state visual evoked potential (SSVEP) in the field of brain-computer interface (BCI). Normally CCA-based methods do not distinguish the different visual feeling of the subject made by the stimuli with different frequencies. To address this issue, this paper proposed a novel method called weighted canonical correlation analysis (WCCA). To recognize the SSVEP frequency, WCCA first employs the standard CCA to obtain its canonical correlation coefficients with each of the reference signals, which correspond to each stimulus frequency. Then each coefficient is modified with a weight, which is associated with each stimulus frequency and obtained by a training procedure. Different from the SSVEP recognition method based on the standard CCA, WCCA uses the weighted coefficients, rather than the original coefficients, to recognize the SSVEP frequency of the testing EEG data. WCCA may be regarded as a generalized CCA, since WCCA is just CCA when each weight value is one. Therefore, by optimizing the weight values, WCCA may outperform CCA. The effectiveness of WCCA was verified by the experiment results.

[1]  Xiaogang Chen,et al.  Filter bank canonical correlation analysis for implementing a high-speed SSVEP-based brain–computer interface , 2015, Journal of neural engineering.

[2]  Xingyu Wang,et al.  SSVEP recognition using common feature analysis in brain–computer interface , 2015, Journal of Neuroscience Methods.

[3]  H. Adeli,et al.  Brain-computer interface technologies: from signal to action , 2013, Reviews in the neurosciences.

[4]  Rajesh P. N. Rao.  Brain-Computer Interfacing: An Introduction , 2010 .

[5]  Xiaorong Gao,et al.  Enhancing the classification accuracy of steady-state visual evoked potential-based brain–computer interfaces using phase constrained canonical correlation analysis , 2011, Journal of neural engineering.

[6]  Changle Zhou,et al.  An SSVEP Recognition Method by Combining Individual Template with CCA , 2019, ICIAI 2019.

[7]  Xingyu Wang,et al.  Frequency Recognition in SSVEP-Based BCI using Multiset Canonical Correlation Analysis , 2013, Int. J. Neural Syst..

[8]  Tzyy-Ping Jung,et al.  High-speed spelling with a noninvasive brain–computer interface , 2015, Proceedings of the National Academy of Sciences.

[9]  H. Hotelling.  Relations Between Two Sets of Variates , 1936 .

[10]  Yiannis Kompatsiaris,et al.  Comparative evaluation of state-of-the-art algorithms for SSVEP-based BCIs , 2016, ArXiv.

[11]  Yu Zhang,et al.  A Novel Multilayer Correlation Maximization Model for Improving CCA-Based Frequency Recognition in SSVEP Brain-Computer Interface , 2017, Int. J. Neural Syst..