Output details

Conditional random fields as classifiers for 3-class motor-imagery brain-computer interfaces

<22> This paper investigated the role of temporal structure information in brain signals in enhancing brain-computer interface (BCI) performance by proposing a novel classifier based on conditional random fields (CRF). Significant performance improvement over non-temporal methods has been demonstrated by extensive experiments. This work was later applied successfully for solving a critical problem in self-paced BCI, i.e., onset detection, by the Essex BCI group, and extended by other BCI researchers (such as Müjdat Çetin at Sabanci University, mcetin@sabanciuniv.edu) to another new approach based on hidden CRF.