Processamento e classificação de sinais de EEG por meio de machine learning para identificação de emoções básicas

Abstract

Electroencephalogram is a method of monitoring electrical activity in the human brain and, with it, it is possible to record waves emitted by this organ. By processing these signals, it is possible to equate and analyze them, seeking to understand the pattern of the individual’s brain activity or even dysfunctions. For this to be possible, it is necessary to apply feature extraction and selection processes to these signals and normalization of the obtained coefficients. The use of machine learning algorithms makes it possible to analyze a larger volume of data and to identify and classify patterns of brain signals. Thus, together with the physiological study of the organ, it is possible to identify emotions by reading and interpreting these waves. In this work, machine learning was used to classify EEG signals and identify 8 basic emotions: satisfied, angry, protected, scared, happy, sad, carefree and surprised. Considering the SVM and KNN classifiers, it is possible to obtain greater accuracy for classifying this type of signal using the KNN.