Enhancing BCI-Based Emotion Recognition Using an Improved Particle Swarm Optimization for Feature Selection.

Area of Science:

• Neuroscience
• Computer Science
• Signal Processing

Background:

• Electroencephalogram (EEG) signals are crucial for emotion recognition.
• Current EEG-based emotion recognition systems suffer from low accuracy and limited real-time application.
• There is a need for improved feature selection algorithms to enhance EEG-based emotion recognition.

Purpose of the Study:

• To propose an improved feature selection algorithm for EEG-based emotion recognition.
• To design an online emotion recognition brain-computer interface (BCI) system using the proposed feature selection method.
• To enhance the accuracy and real-time applicability of emotion recognition from EEG signals.

Main Methods:

• Extracted diverse features from time, frequency, and time-frequency domains of EEG signals.
• Employed a modified particle swarm optimization (PSO) with multi-stage linearly-decreasing inertia weight (MLDW) for feature selection.
• Utilized a support vector machine classifier for emotion type classification.

Main Results:

• Offline experiments on the DEAP dataset achieved an average accuracy of 76.67% for four-class emotion recognition.
• The MLDW-PSO feature selection method demonstrated improved accuracy compared to the latest benchmarks.
• An online two-class emotion recognition system achieved an average accuracy of 89.5% for 10 healthy subjects.

Conclusions:

• The proposed MLDW-PSO feature selection algorithm significantly improves EEG-based emotion recognition accuracy.
• The developed online BCI system shows the practical effectiveness of the MLDW-PSO method for real-time emotion recognition.
• This research advances the field of brain-computer interfaces for affective computing.