ADAPTIVE SUBJECT-SPECIFIC BAYESIAN SPECTRAL FILTERING FOR SINGLE TRIAL EEG CLASSIFICATION

Despite recent advances in signal and information processing, human brain remains the most intriguing signal processing unit with inconceivable abilities to analyze and fuse various multi-modal, streaming signals adaptively in real time. With recent advancements in sensors and computational technologies, brain computer interfacing (BCI) via electroencephalography (EEG) signals have received extensive attention for establishing an alternative form of communication with our brain. In this paper, we propose a subject-specific filtering framework, referred to as the regularized double-band Bayesian (R-B2B) spectral filtering, couples three main feature extraction categories, namely filter-bank solutions, regularized techniques, and optimized Bayesian mechanisms to enhance the classification accuracy by simultaneously taking advantage of the three processing techniques. Furthermore, data collection experiments' are performed to investigate different effects of stimulus on the performance of the proposed R-B2B. In this regard, four different protocols are designed and implemented by introducing visual and voice stimuli. Finally, the paper investigates effects of adaptive trimming of EEG epochs resulting in an adaptive and subject-specific solution. Experimental results show that the proposed R-B2B filter noticeably outperforms its counterparts.