WaveCSP: a robust motor imagery classifier for consumer EEG devices.

摘要: There is an increasing demand for reliable motor imagery (MI) classification algorithms applications in consumer level brain-computer interfacing (BCI). For the practical use, such must be robust to both device limitations and subject variability, which make MI a challenging task. This study proposes methods effect of including limited number electrodes, spatial distribution lower signal quality, variabilities BCI literacy, on performance classification. To mitigate these limitations, we propose machine learning approach, WaveCSP that uses 24 features extracted from EEG signals using wavelet transform common pattern (CSP) filtering techniques. The algorithm shows better terms variability compared existing work. application Physionet database more than 50% 109 subjects achieving accuracy higher 64%. data obtained commercial headset same experimental protocol result up four out five who had prior experience (out total 25 subjects) performing with