Research And Application Of Attention Recognition In Real-time Brain-computer Interface Based On MPI

Attention is a term of psychology, and it is also a focus of Brain scientists and Bioinformaticians. If we have an accessory equipment which recognize human’s attention, we can help ourselves concentrate when in studying or working which can improve our efficiency. With computer, we can develop a system called Brain-Computer Interface (BCI) system to monitor human’s attention. But so far, most BCIs focus on motor function such as mechanical arm and wheelchair controller. Few BCIs focus on attention and fewer BCIs which focus on attention can be used in real-time.This paper will show you what BCI is, the theory and the technologies of BCI. We use a lot of methods and experience as references and develop a BCI system by ourselves to monitor human’s attention during the learning process. The BCI system analyzes human’s EEG (Electroencephalograph) and uses a classifier to get human’s attention state. In the BCI system, we use Wavelet Transform to remove the artifacts in EEG, and after that we extract six features from the EEG, and at last, we classify all the features with a parallelization k-Nearest Neighbor (KNN). BCIs usually need to take some train on each user for some days or some months. So, the data volume of train data may be very big which causing a long latency time in the classifier. In order to apply this BCI to some application systems, we should solve a problem of decreasing the latency time. In the whole BCI system, the classifier is very time consuming. So, in order to decrease the latency time, we replace the KNN with a desterilized KNN classifier based on MPI (Message Passing Interface). The experiments show that our BCI system is effective to the attention-recognition system and can run in real-time.