Adaptive Autoregressive Parameters For A Brain-Computerinterface Experiment

: In an online EEG discrimination task continuous feedback was presented. The EEG was recorded during imagination of left and right hand movement and analyzed with adaptive autoregressive parameters. The parameters discrimination was fed back in form of a rectangular bar on a computer screen over a period of 4 seconds. An online classification result of more than 90% was obtained after a few sessions. I. INTRODUCTION Online analysis and classification of EEG data is the main task in an EEG-based Brain Computer Interface (BCI). Such a BCI transforms specific mental activity (thoughts) into signals to set up a new communication system which can be used by subjects with severe motor disabilities [1, 2 3, 4]. An EEG-based BCI often uses the band-power of selected frequencies, which are classified with a linear threshold [5] or an artificial neural-network [6,7]. random interval feedback CUE one time interval for classification 1 2 3 4 5 6 7 8 t[s] ##R## Figure 1: Scheme of BCI4c paradi..