Biometric system based on electroencephalogram analysis

Searching for new biometric traits is currently a necessity because traditional biometrics such as fingerprint, voice, or face are highly prone to forgery. For this reason, the study of bioelectric signals is outstanding for its potential to develop biometric systems. A motivation for using EEG signals is that they are unique to each person and are much more difficult to replicate than conventional biometrics. The present study is focused on the development of a biometric system based on the analysis of electroencephalograms (EEG). Using six different classifiers: Gaussian Naïve Bayes Classifier (GNB), K-Nearest Neighbors (KNN), Random Forest (RF), AdaBoost (AB), Support Vector Machine (SVM) and Multilayer Perceptron (MLP); a comparison was made between different levels of decomposition of the discrete wavelet transform, used as a preprocessing method. This comparison proved that the level of decomposition does not have a great impact on the overall result of the system. Subsequently, the effect of the recording time of the EEGs on the performance of the system was analyzed, proving that this time is a highly influential factor in overall performance. It is worth mentioning that, during this study, two different data sets were used. Finally, SVM and AB were the best classifiers since they obtained values of sensitivity, specificity, and accuracy greater than 95%.

• Hlavní autor: Carrión Ojeda, Dustin Javier (author)
• Médium: bachelorThesis
• Jazyk: eng
• Vydáno: 2020
• Témata: Biometric; Electroencephalogram; Level of decomposition; Recording time; Classification; Discrete wavelet transform
• On-line přístup: http://repositorio.yachaytech.edu.ec/handle/123456789/245