Application of Pattern Recognition Method in a Linguistic Experiment with Unsupervised Classification

Ali Kamel Issmael Junior, Aline Gesualdi Manhães, José Vicente Calvano

DOI: 10.14209/sbrt.2017.18

Evento: XXXV Simpósio Brasileiro de Telecomunicações e Processamento de Sinais (SBrT2017)

Keywords: ERP EEG Pattern Recognition Linguistics.

Abstract

Event-Related Potentials (ERP) are biological electrical signals synchronized with sensory, cognitive or motor stimuli and measured by electroencephalographs (EEG). ERP technique allows non-invasive analysis of brain functions. Based on the results obtained by Soto [1], this work extracts ERP parameters using EEGLAB® and ERPLAB® tools based on Matlab® software [6], [7], [8], [9]. The result of the research was the obtaining of supervised and unsupervised classification scenarios for the classes proposed in the mentioned experiment and the comparative study and discussion of the classification results found, using the methodology proposed by Webb [2]. This article presents the results obtained with unsupervised classification scenarios only and the supervised classification scenarios will be presented in future. The results achieved accuracies very near from the equiprobability, indicating that the use of unsupervised classifiers approaches considered are not adequate to classify Soto’s data [1]. This study is innovative in the area of Neurolinguistics, since, at least until now, there are no similar previously published works on the subject found in research databases such as: IEEExplorer; Web of Science; Elesevier and Spring. The results open the possibility of analyzing signals from individuals with this ERP methodology associated to Pattern Recognition, with the possible application of this type of analysis in diagnostic tools, assessment of language learning, among others.

Download