EEG-Based Person Identification during Escalating Cognitive Load.

Area of Science:

• Biometrics and Human-Computer Interaction
• Neuroscience and Signal Processing

Background:

• Reliable person identification is crucial for security and privacy.
• Brain signals, specifically electroencephalography (EEG), offer unique patterns for biometric authentication.
• Existing methods require further enhancement for real-world applications.

Purpose of the Study:

• To develop and evaluate a novel biometric identification system using electroencephalography (EEG) signals.
• To investigate the efficacy of increasing cognitive brain load as a basis for person identification.
• To assess the performance of a 1D Convolutional Neural Network (CNN) for classifying EEG patterns.

Main Methods:

• Collected EEG data from 21 subjects undergoing tasks with varying cognitive loads, from relaxation to a serious game.
• Utilized a 1D CNN in MATLAB for time-domain EEG signal pattern recognition and classification.
• Employed 5-fold cross-validation to evaluate identification accuracy for individual tasks and fused tasks.

Main Results:

• Achieved identification accuracies exceeding 99% for individual cognitive tasks.
• Reached accuracies of over 98% when fusing data from multiple tasks.
• Demonstrated the feasibility of reducing the number of EEG channels without significant performance degradation.

Conclusions:

• The proposed EEG-based biometric identification method, utilizing cognitive load variations, is highly accurate and effective.
• The 1D CNN approach provides robust classification of EEG patterns for person authentication.
• This brainwave-based identification system shows significant promise for real-world security applications.