Stability of Human EEG Patterns in Different Tasks: The Person Authentication Problem
- PDF / 493,447 Bytes
- 7 Pages / 594 x 792 pts Page_size
- 108 Downloads / 163 Views
Stability of Human EEG Patterns in Different Tasks: The Person Authentication Problem N. N. Lebedeva and E. D. Karimova
UDC 612.821
Translated from Zhurnal Vysshei Nervnoi Deyatel’nosti imeni I. P. Pavlova, Vol. 70, No. 1, pp, 40–49, January–February, 2020. Original article submitted April 15, 2019. Revised version received July 26, 2019. Accepted September 16, 2019. This article addresses the main problems in person authentication using the EEG. This area is currently under active development due to advances in virtual spaces and seeking new methods for user recognition in different internet platforms. One task that needs to be solved is that of identifying stable EEG measures and patterns which might be used to perform reliable recognition of people over long time intervals. The second question considered here is that of selecting tasks for EEG recording protocols. In the present study, subjects’ EEG traces were recorded at rest and on performance of various motor tasks five times over three months and the stability of the different patterns was then compared. The results showed that the most stable was the α-rhythm pattern in the resting state with the eyes closed, with minimal values of the coefficient of variation of the α rhythm but strong within-group spreads. Of the active tests, the most stable indicators were obtained on observing motor actions and the least stable patterns were seen on performance. Writing with a pen was the action characterized by the lowest stability of EEG measures. Keywords: EEG, stability, identification, personal authentication.
cephalograph (EEG) has unrivaled universality for different tasks and absolute uniqueness for each person, thus minimizing the risk of fraud. The first identification based on individual EEG characteristics recorded in the resting state was demonstrated in 1999 [Poulos et al., 1990a; Poulos et al., 1999b; Poulos et al., 2002]. However, most efforts in working with individual EEG patterns have been focused on the “brain–computer interface” problem, although it was already understood that quite stable EEG patterns can be recorded on hand or finger movements. The main problem with the widespread use and commercialization of personal recognition based on the EEG is not in mobile rapid reading or recording of the EEG – one study [Chuang et al., 2013] demonstrated that recording of a single EEG channel with dry electrodes was enough. The main problem was extraction of stable patterns and EEG indicators which allowed accurate and reliable identification regardless of the affective and physical state of the person and also regardless of the time since the last recording. An authentication or identification system must be able to recognize recorded clients even when they return days,
Introduction. The contemporary world and society are developing towards virtual space: social network accounts, online banking, cloud data storage, automobile rentals, online shops, and internet sales. All these services are mobile and more accessible, though on the other hand they
Data Loading...
