Biometric identification and authentication via resting state electroencephalography (EEG) alpha frequency

Biometric identification and authentication are systems that enable an individual to be identified and distinguished from other individuals by examining their physical and behavioral characteristics. The most common biological signals used in biometric identification and authentication studies are fingerprint, facial recognition and iris recognition. In order to overcome the disadvantages of these systems, such as not being able to detect liveness, being easy to mislead, and not being able to get the same quality data, a different biometric approach, brain signals, has been used. Electroencephalogram (EEG) is a method that allows recording the electrical activity in the brain with the help of electrodes attached to the scalp. EEG alpha signals, which have high amplitude values and occur mostly in the parietal and occipital regions of the brain at rest, are in the frequency range of 8-13 Hz. In this study, biometric identification and authentication applications were carried out using resting state EEG alpha frequency unique brain signals. In the data set, EEG signals of twenty people were obtained using a sampling frequency of 512 Hz per second. Machine and deep learning models have been applied to EEG alpha frequency signals for biometric identification and authentication. The most common performance metrics were examined to evaluate the performance of biometric identification and authentication studies. In the biometric identification study, the performance of different sampling deep learning models was investigated and compared. In the biometric authentication study, performance metrics were evaluated using only two channel data. The results show that both identification and authentication studies have high prediction performance and accuracy values. In this study, a low-cost, easy-to-use and reliable biometric identification and authentication system is proposed.