Common feature analysis for recognizing steady-state visual evoked potential in brain-computer interface

Canonical correlation analysis (CCA) has been successfully applied to steadystate visual evoked potential (SSVEP) recognition for brain-computer interface (BCI) application and outperforms the traditional power spectral density analysis through multichannel detection with resorting to the pre-constructed reference signals of sine-cosine waves. However, the CCA method is like to encounter overfitting in using a short time window length since the reference signals include no features from training data. We consider that a group of electroencephalogram (EEG) data trials recorded at a certain stimulus frequency on a same subject should share some common features that may bear the real SSVEP characteristics. This study therefore proposes a common feature analysis (CFA)-based method to exploit the latent common features as natural reference signals in using correlation analysis for SSVEP recognition. Effectiveness of the CFA method is validated with EEG data recorded from ten healthy subjects, in contrast to CCA and a multiway extension of CCA (MCCA). Experimental results indicate that the proposed CFA method significantly outperformed the CCA and the MCCA methods for SSVEP recognition in using a short time window length (i.e., less than 1 s). This superiority suggests that the CFA method is promising for the development of a high-speed SSVEP-based BCI.