Análise comparativa de classificadores de padrões para classificação de tarefas mentais

Abstract

The electroencephalogram is a technique that uses electrodes positioned on the surface of the head to obtain the biological signals present in the brain in a graphic way. With these charts, it is possible to diagnose and monitor diseases that affect the nervous system. In addition, the electroencephalogram allows monitoring what happens to brain signals when performing mental tasks, such as imagining the movement of the upper limbs. These mental tasks can be used for applications such as brain computer interfaces, for which the signals must be acquired in real time through the electroencephalogram and classified by pattern classifiers. For this purpose, electroencephalogram signals obtained through experiments were used to test the performance of the k-nearest neighbor classifiers and artificial neural networks to classify the activities present in these signals. And with that, compare the results obtained through the classifiers to define which classifier had the best performance in this activity.