SSVEP recognition using multivariate linear regression for brain computer interface

Until now, the canonical correlation analysis (CCA)-based method has been most widely applied to steady-state visual evoked potential (SSVEP). Artificial sine-cosine signals are used as the original references in the CCA method, which could hardly reflect the real SSVEP features buried in electroencephalogram (EEG). In this study, we use principal component analysis (PCA) to extract EEG features multivariate linear regression (MLR) is implemented on EEG and the specific sample labels. Experimental results show that the proposed MLR method outperformed other two competing methods for SSVEP recognition, especially in short time window.