A Multi-Agent System for Improving Electroencephalographic Data Classification Accuracy

Electroencephalographic (EEG) devices are utilized to measure the electrical activity of the human brain cost-effectively. In this technology, an electrical potential available on the scalp is measured. Special kind of sensors called electrodes are positioned on the scalp following international standards. One of the key benefits of the Electroencephalography is, the detectability of some brain disorders such as Epileptic Seizure. In addition to the medicinal usage, the EEG technology is often preferred by Brain Machine Interfacing (BMI) or Brain-Computer Interfacing (BCI) researchers to recognize a patient's intentions. The objective is to control computers or machines according to the user's intentions. In other words, BCI / BMI is an alternative hands-free Human-Computer Interaction (HCI) system which replaces the typical input devices such as a mouse or keyboard. In most BMI or BCI applications, a noninvasive EEG data acquisition approach is followed, using a consumer-grade EEG device. Such a device is equipped with only a few electrodes, causing a major drawback, limited accuracy (typically less than 70%). The only remedy for this issue is, improving the accuracy of the EEG data classifier, the computational algorithm to recognize the user intentions. In this paper, the applicability of a Multi-Agent System for EEG data classification is discussed, which has confirmed its competency in improving the accuracy by 17%, approximately.