Classification strategies for a single-trial binary Brain Computer Interface based on remembering unpleasant odors

A Brain Computer Interface (BCI) is a useful instrument to support human communication. In recent years, BCI systems have been frequently implemented by using EEG. Regarding the communication paradigm used, there exists a very large number of strategies and, recently, the remembering of unpleasant odors has been also defined. However, the quality of the signals collected by this last paradigm is very poor, due to the absence of a real stimulus (the stimulus consists in remembering a disgusting situation). For this reason, a crucial node is the choice of a very efficient classification algorithm to improve the accuracy of the BCI. The present paper describes a and compares classification strategies for such type of BCI systems. The proposed methods and the experimental setup are described and experimental measurements are presented and discussed.