On the brain activity misclassification: SSVEP-based analysis

This work deals to present some problems related to the analysis of the brain activity of a user when he detects an error in the classification or the interpretation of his previous intention. This error can be detected after each correct or incorrect classification of a brain activity. Our study is based on the visual stimulation using SSVEP. Electrodes have to be installed on the scalp of the user to measure EEGs signals reflecting cerebral activity. The main idea of this work deals with the automatic detection of each classification error without any a priori knowledge about the shape of EEG signals. The only hypothesis considered is that the user's brain activities when observing a correct classification are not similar to those corresponding to incorrect classification. On the other hand, the proposed study concerning the recognition procedure is based on the existence of similarities among the user brain activities during his observation of several classification errors.