Frequency bands based on EEG typing for biometric authentication

• Published in: AIP conference proceedings Vol. 2339; no. 1
• Main Authors: Amalina, I., Saidatul, A., Fook, C.Y.
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Melville American Institute of Physics 03.05.2021
• Subjects: Bandpass filters; Biometrics; Classifiers; Electroencephalography; Engineering education; Feature extraction; Frequencies; Gait; Independent component analysis; Notch filters
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0044545

Biometric authentication plays a great role in determining the identity of a person used by researchers all over the world. In other word, biometrics implies all innovative procedures utilized to confirm or distinguish people depending on their physical and/ or behavioral characteristics. Even so, the single biometric traits, such as face, iris, gait, touch gestures, images and fingerprint, mostly fails to maintain the security requirements of many applications with high population of data. Therefore, the aim of this research is to examine the electroencephalogram (EEG) signals during typing task performed by five engineering students (right-handed) to extract the useful information from a person identity. All subjects were required to perform two sets of typing tasks for 3 minutes. The subjects were asked to carry out familiar typing task (typing first and last name) and unfamiliar (typing random names twice for 3 minutes) that was provided. EEG signals were recorded by using EEGOTM sports device (ANT Neuro, Enschede, The Netherlands). This study applied Independent Component Analysis (ICA), notch filter and Butterworth Bandpass filter to remove 50Hz powerline artefacts and proposed the implementation of linear feature such as mean, median, standard deviation and variance using Burg method and Welch Method. The extracted entropies features vector are used as an input to k-Nearest Neighbour (k-NN) and Random Forest (RF) classifier. As a conclusion, the k-NN and RF classifier giving promising accuracy which are 98.91% and 99.89% respectively for beta band.