EEG Channel Selection Based User Identification via Improved Flower Pollination Algorithm.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Machine Learning

Background:

• Electroencephalogram (EEG) signals offer unique features for robust user identification and defense against spoofing attacks.
• Accurate user identification via EEG is crucial, but selecting the optimal subset of electrodes for signal capture remains a significant challenge.
• Existing electrode selection methods often require extensive computational resources or fail to achieve optimal performance.

Purpose of the Study:

• To introduce a novel, efficient algorithm for selecting the most representative EEG electrodes for enhanced user identification.
• To address the challenge of optimal electrode selection in EEG-based biometrics.
• To improve the accuracy and efficiency of EEG-based identification systems.

Main Methods:

• Development of a hybrid optimization algorithm, FPAβ-hc, combining Flower Pollination Algorithm and β-Hill Climbing.
• Formulation of electrode selection as an optimization task.
• Evaluation of the FPAβ-hc algorithm on a standard EEG motor imagery dataset.

Main Results:

• The FPAβ-hc algorithm successfully identified the most representative electrodes.
• The proposed method achieved higher accuracy compared to seven other existing methods.
• FPAβ-hc utilized less than half the number of electrodes required by other methods, demonstrating significant efficiency.

Conclusions:

• The FPAβ-hc algorithm presents a highly effective and efficient approach for EEG electrode selection in user identification.
• This method significantly enhances identification accuracy while reducing the number of electrodes, paving the way for more practical EEG biometric systems.
• The findings suggest that optimized electrode selection is critical for maximizing the potential of EEG in security applications.