Multi-subject spatial filtering in brain-computer interfaces
暂无分享,去创建一个
Brain-computer interfaces (BCI) present new communication alternatives for severely disabled people. Although these BCI systems exist in many flavors, the system we focus on is based on the imagination of hand and/or foot movements. The interface then tries to detect the correct imagined movement from the subject’s brain signals. These kind of systems commonly use the common spatial pattern filter (CSP) as preprocessing step before features are extracted from the EEG signals and classified as one of the mental movement tasks (also referred to as classes). The CSP method is a supervised algorithm and therefore needs subject specific training data for calibration, which is very time consuming to collect. Instead of letting all that data and effort go to waste, we could use the data of other subjects to further improve the BCI performance for new subjects. This problem setting is often encountered in multitask learning, from which we borrow some ideas and apply it to the preprocessing phase.
[1] G. Pfurtscheller,et al. Designing optimal spatial filters for single-trial EEG classification in a movement task , 1999, Clinical Neurophysiology.