An Automatic Optimum Data selection Method For EEG-based Brain-computer Interface

An electroencephalogram (EEG) based brain-computer interface (BCI) is aimed at developing a system that can support communication possibilities for patients with severe neuromuscular disabilities through EEG pattern recognition and classification. Previously many parametric modeling techniques for EEG analysis have been developed and improved upon. For this work we analyzed five parameters on seven subjects to study their influence on brain computer interface (BCI) classification. Our study shows that these parameters greatly influence classification accuracy with subject dependent parameters. This suggests that the parameter selection process should be analyzed further when building models.