Bci-based emotion recognition (reconocimiento de emociones basado en interfaces bci).

Abstract

This study aims to contribute to the field of emotion recognition using brain-computer interfaces for EEG signal collection. In this thesis is conducted an investigation on recognition of emotions using two data sets of EEG signals generated by us, with participants subjected to two types of methods of elicitation of emotions: activities in the Stock Market and the online poker game. The data sets were generated as part of the study to investigate active methods to provoke emotions and offer a new perspective, in which EEG signals are collected using two different devices with 8 and 14 electrodes respectively, applied to participants who perform trading activities and online poker game. These activities were chosen because they are essentially unpredictable, playful and because they are capable of producing strong emotions in the participants. In addition, this form of elicitation of emotions is interactive and dynamic in contrast to the predominant methods in which only passive stimulation with images, sounds or films is used. Machine learning systems are used consisting of the following phases: pre-processing, feature extraction, feature selection, machine learning algorithms (training and testing) and performance evaluation. With them, the recognition of emotions is carried out in the valence – arousal plane. The results obtained with the classification algorithms used were satisfactory, comparable or better than those of the state of the art.