A Framework for Robust Person Identification Using Brain Signals

Biometric authentication is pivotal in identifying individuals with unique physiological or behavioral characteristics. General recognition methods, such as fingerprint, voice, iris, and face recognition, are widely used but have significant flaws. These can be sensitive to spoofing, raise privacy concerns, and often struggle in certain environments. To fix these shortcomings, we proposed a novel biometric method: Electroencephalogram (EEG) authentication. Electroencephalogram (EEG) technology measures brainwave activity through electrodes and is known for its reliability, resistance to forgery, and inherent uniqueness, similar to fingerprints. EEG is particularly significant for liveness detection, making it a strong candidate for robust biometric authentication in high-security applications. This study utilizes a publicly available dataset consisting of EEG data from 109 subjects. The raw data is first scaled and then analyzed using various classifiers, such as <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\kappa$</tex>-nearest neighbors (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\kappa$</tex>-NN), Auto-Encoder with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\kappa$</tex>-NN, and Convolutional Neural Networks (CNN). The model's performance is evaluated under four different conditions based on the subjects' activities, with the CNN achieving an authentication accuracy of 92%.