Brainprint: Assessing the uniqueness, collectability, and permanence of a novel method for ERP biometrics

摘要: Abstract The human brain continually generates electrical potentials representing neural communication. These can be measured at the scalp, and constitute electroencephalogram (EEG). When EEG is time-locked to stimulation – such as presentation of a word averaged over many presentations, Event-Related Potential (ERP) obtained. functional characteristics components ERP are well understood, some represent processing that may differ uniquely from individual individual—such N400 component, which represents access semantic network. We applied several pattern classifiers ERPs response individuals stream text designed idiosyncratically familiar different individuals. Results indicate there robustly identifiable features enable labeling belonging with accuracy reliably above chance (in range 82–97%). Further, these stable time, indicated by continued accurate identification after lag up six months. Even better, high degree achieved in all cases was use only 3 electrodes on scalp—the minimal possible number acquire clean data.