Brainprint Recognition Based on the Stable SSVEP Space-Frequency Energy Distribution

Area of Science:

• Neuroscience
• Biometrics
• Signal Processing

Background:

• Brainprint recognition using electroencephalography (EEG) signals is gaining traction in information security.
• The specific components contributing to the uniqueness of individual EEG signals remain largely unidentified.
• Existing methods lack interpretability regarding the source of brain signal distinctiveness.

Purpose of the Study:

• To propose an interpretable biomarker for EEG biometric identification based on steady-state visual evoked potentials (SSVEP).
• To investigate the spatial and frequency distribution properties of SSVEP components for subject separability.
• To develop a robust brainprint recognition system utilizing these unique SSVEP features.

Main Methods:

• Pure SSVEP components were isolated from EEG data using a point-position equivalent reconstruction (PPER) method.
• Spatial and frequency distribution properties of SSVEP components were analyzed.
• Uniform Manifold Approximation and Projection (UMAP) was employed for dimensionality reduction, followed by Long Short-Term Memory (LSTM) network for classification.

Main Results:

• The proposed method successfully identified unique SSVEP components, demonstrating significant subject separability in a 2-dimensional feature space.
• The Long Short-Term Memory (LSTM) network achieved an average recognition accuracy of 98.33% on the SSVEP benchmark dataset.
• The space-frequency energy feature of SSVEP was confirmed as an effective and interpretable biomarker.

Conclusions:

• The space-frequency energy feature of SSVEP serves as a highly effective and interpretable biomarker for brainprint recognition.
• This research enhances the understanding of individual EEG signal uniqueness, paving the way for advanced personal identification systems.
• The findings support the practical application of SSVEP-based biometrics in information security.