Abstract—For realizing continuous authentication of users, we have studied to use an electroencephalogram (EEG) evoked by ultrasound as biometrics. Users are presented only the ultrasound of their memorable music and verified whether genuine or not using the induced components of EEG. In our previous studies, the verification error rate of 0 % was achieved using multiple quantities in EEG as individual features and a support vector machine (SVM) as a verification method; however, it required a large amount of computation for processing SVM models. Thus, we reduce the number of SVM models by applying two selection methods of features and electrodes, which have been previously introduced. Furthermore, we examine the usage rates of features and electrodes in the reduced SVM models. By using only the electrodes with high usage rates, the verification error rate of 0 % is guaranteed with a small amount of computation. 

Index Terms—biometrics, electroencephalogram, ultrasound, support vector machine, computational amount reduction

Cite: Isao Nakanishi and Kouki Kawakami, "Reduction of Computational Amount in Person Verification Based on SVM Using Evoked Brain Wave by Ultrasound," International Journal of Signal Processing Systems, Vol. 11, No. 4, pp. 17-24, December 2023. doi: 10.18178/ijsps.11.4.17-24

Copyright © 2023 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.