Deep learning-based self-induced emotion recognition using EEG.

Area of Science:

• Neuroscience
• Computer Science
• Signal Processing

Background:

• Emotion recognition from electroencephalogram (EEG) signals is crucial for understanding human affective states.
• Deep learning models offer advanced feature extraction for improved EEG-based emotion classification.
• Classifying self-induced emotions from high-density EEG presents challenges like computational complexity and channel redundancy.

Purpose of the Study:

• To investigate the effectiveness of a channel selection method for deep learning-based emotion recognition from EEG signals.
• To address the challenges of high computational complexity and low accuracy in high-density EEG classification.
• To evaluate the performance of a kurtosis-based channel selection method for classifying self-induced emotions.

Main Methods:

• Utilized deep learning for automatic feature extraction from electroencephalogram (EEG) signals.
• Implemented a channel selection method based on signal statistics to reduce redundant channels.
• Experimentally evaluated the kurtosis-based channel selection method for classifying self-induced emotions on valence and arousal scales.

Main Results:

• The proposed channel selection method reduced computational complexity by 89% while maintaining classification accuracy.
• The kurtosis-based channel selection achieved the highest accuracy, reaching 79.03% for valence and 79.36% for arousal.
• The framework demonstrated superior performance compared to conventional methods despite using fewer channels.

Conclusions:

• Channel selection is an effective strategy to optimize deep learning models for EEG-based emotion recognition.
• The proposed method enhances efficiency and accuracy, making EEG signal analysis more practical for real-world applications.
• This approach facilitates the development of more accessible and computationally feasible emotion recognition systems.