Motif Synchronization and Space-Time Recurrences for Biometry from Electroencephalography Data: A Proof-of-Concept

The electroencephalography (EEG) technique has the capability of identifying individual traits. Previous work has already used functional connectivity (FC) features obtained from EEG data for biometric purposes. In this work, we explored two FC methods not yet used in this context: motifs synchronization and space-time recurrences. Fifty subjects with two resting-state EEG acquisitions (one with eyes open and another with eyes closed) were included in the study. FC matrices for 1 s and 5 s epochs were computed for each acquisition. Subject’s identification was sought by comparing the FC matrices from both acquisitions using the Pearson correlation coefficient. The motifs method achieved 48% accuracy for both epoch sizes, and the space-time recurrences achieved 36% and 38% accuracies for 1 s and 1 s epochs respectively. Although the accuracies were low, they were well above the 2% chance level. Also, unlike other similar studies, the comparison was made between signals acquired in different conditions. In general, the obtained low accuracies illustrate the challenging problem of performing biometry from EEG and the need for further adjustments in the feature extraction and classification stages.