Related Experiment Video
Updated: Jan 10, 2026
11:15
Applications of EEG Neuroimaging Data: Event-related Potentials, Spectral Power, and Multiscale Entropy
Published on: June 27, 2013
34.3K
Evaluation of entropy features and classifier performance in person authentication using resting-state EEG
Renyu Yang1,2, Ling Zhang3, Yuanmei Peng4
1School of Big Data and Artificial Intelligence, Guangdong University of Finance & Economics, Guangzhou, China.
Frontiers in Neuroscience
|November 20, 2025
Summary
Resting-state electroencephalogram (EEG) biometrics achieve high accuracy with optimal electrode and entropy feature selection. A 9-electrode system maintains performance, enabling efficient portable biometric devices.
Area of Science:
- Biometrics and Human-Computer Interaction
- Neuroscience and Signal Processing
- Cybersecurity and Authentication
Background:
- Resting-state electroencephalogram (EEG) offers inherent liveness detection and spoofing resistance for biometric systems.
- Conventional biometric systems face vulnerabilities; EEG presents a promising alternative.
- Challenges in EEG biometrics include optimizing electrode configuration, feature extraction, and classifier generalization for accuracy, robustness, and hardware efficiency.
Purpose of the Study:
- To systematically evaluate entropy measures and classifiers for person authentication using EEG.
- To quantify the impact of electrode selection and feature-classifier pairing on biometric performance.
- To assess the feasibility of portable EEG biometric devices with reduced hardware requirements.
Main Methods:
- Evaluation of thirteen entropy measures (e.g., spectral entropy (SpEn), sample entropy (SaEn)) and six classifiers (e.g., Quadratic Discriminant Analysis (QDA)).
- Utilized 32-channel EEG recordings from 26 healthy participants.
- Employed rigorous leave-one-out cross-validation (LOOCV) to assess performance.
Main Results:
- Quadratic Discriminant Analysis (QDA) achieved peak accuracy of 96.8% with 30 electrodes.
- A 9-electrode configuration retained 96.1% accuracy, demonstrating robust performance with reduced hardware.
- Spectral entropy (SpEn) showed superior biometric discriminability over sample entropy (SaEn) by 13.8 percentage points.
Conclusions:
- The findings support the development of portable EEG biometric devices.
- Entropy features demonstrate scalability for enhanced biometric system design.
- Optimized electrode selection and feature-classifier combinations are crucial for robust EEG biometrics.
