2024-01-052024-01-052023-12-18SOUSA, Sávio Milhomens de. Apresentação do artigo elaborado em 2023 por meio da Proeg nº 01/2023: aplicação da análise de correlação canônica em sistemas ICM baseados em SSVEP. Orientador: Cleison Daniel Silva. 2023. 22, [2] f. Trabalho de Curso (Bacharelado em Engenharia Elétrica) – Faculdade de Engenharia Elétrica, Campus Universitário de Tucuruí, Universidade Federal do Pará, Tucuruí, 2023. Disponível em: https://bdm.ufpa.br/jspui/handle/prefix/6497. Acesso em:.https://bdm.ufpa.br/jspui/handle/prefix/6497This work addresses the presentation of the article entitled “Application of Canonical Correlation Analysis in SSVEP-based BCI Systems”, developed during the period from April 2023 to September 2023, during the research project, under the guidance of Professor Dr. Cleison Daniel Silva and presented at the “III Escola Regional de Aprendizado de Máquina e Inteligência Artificial Norte 2”. The BCI (Brain-Machine Interface) system is a technology capable of performing communications between humans and machines through brain activity in response to visual, imaginary, or somatosensory stimuli. This activity is acquired through methods such as electroencephalogram (EEG), processed, and converted into command signals. The study of the article focuses on information processing using the Canonical Correlation Analysis (CCA) method to assist in the classification of signals in BCI systems based on SSVEP (Steady-State Visually Evoked Potential). As a methodology, five CCA method approaches were performed in a Python environment, using the same data set and the same signal processing and classification technique, changing only the way the data is treated in the method. The data used comes from a public domain repository containing EEG signals from four subjects in the presence of SSVEP stimuli at frequencies of 8 Hz, 14 Hz and 28 Hz. In addition to CCA, the periodogram was used as a technique to maximize the extraction of signal characteristics resulting from the application of the method. In the classification step, the signals were grouped into three binary combinations between the stimulus frequencies, and for each pair, Linear Discriminant Analysis (LDA) was applied. Finally, the accuracy of the classifier was used as a parameter for discussions and conclusions of each approach. In general, it was noted that the results vary between individuals in a range of 38% to 100% accuracy. From the construction logic and the results of approach E, it is concluded that it is suitable for application in a real system.Acesso AbertoICMSSVEPProcessamento de dadosCCABCIData processingCNPQ::ENGENHARIASCNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAOTrabalho de Curso - Graduação - Artigo