Identificação de atividades cognitivas a partir de modelos de aprendizado de máquina aplicados ao processamento de sinais de Eletroencefalograma (EEG)

• Ano de defesa: 2022
• Autor(a) principal: SILVA, Juliana Mycaelle Oliveira
• Orientador(a): BARROS FILHO, Allan Kardec Duailibe
• Banca de defesa: BARROS FILHO , Allan Kardec Duailibe , SOUSA, Gean Carlos Lopes de , ROCHA, Priscila Lima
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Maranhão
• Programa de Pós-Graduação: PROGRAMA DE PÓS-GRADUAÇÃO EM ENGENHARIA DE ELETRICIDADE/CCET
• Departamento: DEPARTAMENTO DE ENGENHARIA DA ELETRICIDADE/CCET
• País: Brasil
• Palavras-chave em Português: Eletroencefalograma; EEG; Aprendizado de máquinas; Inteligência Artificial.
• Palavras-chave em Inglês: Electroencephalogram; EEG; Machine Learning; Artificial Intelligence.
• Área do conhecimento CNPq: Ciência da Computação
• Link de acesso: https://tedebc.ufma.br/jspui/handle/tede/4210

The different types of activities performed by the subject, such as reading; lis- ten to music; to dance; among others, they lead to the activation of brain regions. Among these activities, cognitive activities are associated with an activation of brain regions related to learning, such as the frontal lobe regions (Superior Frontal, Pre- cuneus, among others). Several studies have been developed to relate cognitive activities and associated brain regions. This type of study is important in unders- tanding the functionality and connectivity of the brain and this knowledge can help to diagnose abnormalities in its functioning. This study aims to develop a classifier model of brain activities during the development of a cognitive activity within the three categories of activity: Video game, music or mathematics. For this, we used the Electroencephalogram (EEG) signals collected in two public databases and using the technique of estimation of brain sources, the anatomical regions related to each of the activities were estimated. After this determination, a model based on ma- chine learning was trained and tested that classifies the type of activity performed according to the activity categories. From the obtained results, we can highlight the en-cephalic activities classifier model elaborated with an accuracy of 99.9%.