Emotion recognition from single-trial EEG based on kernel Fisher's emotion pattern and imbalanced quasiconformal...

Area of Science:

• Neuroscience
• Affective Computing
• Biomedical Engineering

Background:

• Electroencephalogram-based emotion recognition (EEG-ER) is crucial for healthcare and brain-computer interfaces.
• Recognizing emotions in a continuous, bi-dimensional space from single EEG trials is challenging.

Purpose of the Study:

• To develop and validate a robust three-layer scheme for single-trial EEG-ER.
• To improve the accuracy of emotion recognition in terms of valence and arousal.

Main Methods:

• Extracted spectral powers from multi-channel single-trial EEG signals.
• Applied kernel Fisher's discriminant analysis to derive kernel Fisher's emotion patterns (KFEP).
• Utilized an imbalanced quasiconformal kernel support vector machine (IQK-SVM) for classification.

Main Results:

• The proposed KFEP feature demonstrated superior performance over other spectral power features.
• IQK-SVM achieved higher EEG-ER accuracy compared to traditional SVM.
• The system attained high classification accuracies: 82.68% for valence and 84.79% for arousal.

Conclusions:

• The proposed three-layer EEG-ER scheme effectively recognizes emotions from single-trial EEG data.
• The KFEP feature and IQK-SVM classifier offer significant improvements in EEG-ER performance.
• This approach holds promise for advanced applications in affective computing and BCI.