Feature Selection for Motor Imagery EEG Classification Based on Firefly Algorithm and Learning Automata.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Motor Imagery (MI) electroencephalography (EEG) offers non-invasive brain-computer interfaces (BCIs) with high temporal resolution.
• High-dimensional features in MI EEG signals pose challenges for classification accuracy and computational efficiency.
• Existing feature selection methods like the firefly algorithm (FA) can be trapped in local optima.

Purpose of the Study:

• To develop an optimized feature selection method for MI EEG signals.
• To address the limitations of the firefly algorithm in feature selection.
• To improve the classification accuracy of MI EEG signals for BCI applications.

Main Methods:

• A hybrid approach combining the firefly algorithm (FA) with learning automata (LA) for optimized feature selection.
• Utilizing Common Spatial Pattern (CSP) and Local Characteristic-Scale Decomposition (LCD) for high-dimensional feature extraction.
• Employing the Spectral Regression Discriminant Analysis (SRDA) classifier for signal classification.

Main Results:

• The proposed FA-LA method effectively reduces redundant features in MI EEG data.
• Improved classification accuracy for MI EEG signals compared to genetic and particle swarm optimization algorithms.
• Validation using benchmark BCI competition data and real-time experimental data.

Conclusions:

• The combined FA-LA method offers a robust solution for feature selection in MI EEG.
• The proposed method enhances the performance of BCI systems by improving classification accuracy.
• Demonstrated feasibility for real-time BCI applications through system implementation.