A Survey on Brain Biometrics

Abstract

Brainwaves, which reflect brain electrical activity and have been studied for a long time in the domain of cognitive neuroscience, have recently been proposed as a promising biometric approach due to their unique advantages of confidentiality, resistance to spoofing/circumvention, sensitivity to emotional and mental state, continuous nature, and cancelability. Recent research efforts have explored many possible ways of using brain biometrics and demonstrated that they are a promising candidate for more robust and secure personal identification and authentication. Although existing research on brain biometrics has obtained some intriguing insights, much work is still necessary to achieve a reliable ready-to-deploy brain biometric system. This article aims to provide a detailed survey of the current literature and outline the scientific work conducted on brain biometric systems. It provides an up-to-date review of state-of-the-art acquisition, collection, processing, and analysis of brainwave signals, publicly available databases, feature extraction and selection, and classifiers. Furthermore, it highlights some of the emerging open research problems for brain biometrics, including multimodality, security, permanence, and stability. CCS Concepts: • Security and privacy → Biometrics; • Computing methodologies → Machine learning approaches; • Applied computing → Psychology;

EEG fingerprinting: Subject-specific signature based on the aperiodic component of power spectrum

“…Nevertheless, it is relevant to notice that our work is especially focused on relative (and not absolute) performance, since we present the results as a strict comparison between different characteristics of the EEG spectrum (namely, aperiodic and periodic components). Furthermore, the effects induced by different sessions on subject identification in EEG have been widely investigated and clarified (Gui et al 2019) and out of the scope of the present study.…”

Biometric From Surface Electromyogram (sEMG): Feasibility of User Verification and Identification Based on Gesture Recognition

“…With 13 samples in postprocessing, the highest EER values were 6.8% (TA), and the EER of eight gestures, TIFE, TLFE, TFE, WF, WE, FS, FP, and HC were <1.0%. For reference, to the best of our knowledge, the performance of ECG and EEG, which was calculated based on the same data settings as this study (training and testing data from the same day), was mainly in the range from 0.1 to 5% (Pinto et al, 2018) and 1 to 20% (Campisi and Rocca, 2014;Fraschini et al, 2015;Gui et al, 2019), respectively, depending on the algorithms and electrode settings (number and positions) adopted.…”

“…Compared to the three common traits mentioned above, the hidden nature of electrical biosignals makes them harder to capture, synthesize, and imitate, and the inherent liveness nature ensures their robustness in distinguishing the artifacts from the real biological targets. There are a number of studies investigating the use of ECG and EEG signals as biometric traits (Campisi and Rocca, 2014;Pinto et al, 2018;Gui et al, 2019). Other than ECG and EEG, the surface electromyogram (sEMG) signal is another typical electrical biosignal, representing the electrical activities of the muscles.…”

Multi-objective optimization for EEG channel selection and accurate intruder detection in an EEG-based subject identification system

“…In general, resting-state potentials and ERPs have been shown to be good candidates for a new biometric system for which there are several different state-of-the-art approaches 6,[9][10][11][12] , with the localization of the relevant channels differing, depending on the paradigm. Here, we used the ERP approach to stimulate the EEG signals.…”