•  
  •  
 

Abstract

Biometric systems are a continuouslyevolving and promising technologicaldomain that can be used in automatic systems for the unique and efficient identificationandauthenticationof individualswithout necessitating usersto carry or remember any physical tokens or passwords, in contrast totraditional methods such aspassword IDs. Biometricsare biological measurements or physical characteristics that can be used to ascertain and validate the identityof individuals.Recently, considerable interest has emergedin exploiting brain activity as a biometric identifier in automatic recognition systems, particularly focusing on data acquired through electroencephalography (EEG). Multiple research endeavorshaveindeed confirmed the presence of discriminative characteristics within brain signals recorded while performing specific cognitivetasks.However,EEG signals are inherently complex due totheir nonstationary and high-dimensionalproperties,thus demanding carefulconsideration duringboth the feature extraction and classification processes. This study applieda hybridization technique integratinga pre-trained convolutional neural network(CNN) with a classical classifier and the short-time Fourier transform (STFT)spectrum.We used a hybrid modeltodecodetwo-class motor imagery(MI)signalsformobile biometric authentication tasks,which include subject identification andlock and unlock classification. To thispurpose, nine potentialclassifiers (mostlyclassification algorithms)were utilized to build nine distinct hybrid models,with the ultimate goal of selecting the most effectiveone.Practically, sixexperiments were conducted in the experimental part of this study. The first experiment aimsto develop a hybrid modelfor biometric authentication tasks. To do this, nine possible classifiers (mostly classification algorithms) were usedto build nine hybrid models. It can be seen thatthe RF-VGG model achieved better performance compared with other models. Therefore, it was chosen to be utilized for mobile biometric authentication. The fourth experiment is to apply the RF-VGG model for doing the lock and unlock classification process,and their mean accuracy is 97.50%. Consequently, the fifth experiment was conducted to validate the RF-VGG model for the lock and unlock task,and their mean accuracy was 97.40%. Practically, the sixth experiment was to verify the RF-VGGmodel for the lock and unlock task over another dataset (unseendata),andtheir accuracy is 94.4%. Itcan be deduced that the hybrid model appraises the capability ofdecoding the MI signal for the left and right hand. Therefore, the RF-VGG model can contribute to the BCI-MIcommunity by facilitating the deploymentofthe mobile biometric authenticationtask for (the subject identificationand the lock and unlock classification)

Download

Share

COinS