A Brain Network Analysis-Based Double Way Deep Neural Network for Emotion Recognition.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Signal Processing

Background:

• Accurate human emotion recognition is crucial for various applications.
• Existing models often struggle with the complexity and nuances of emotional states.

Purpose of the Study:

• To develop a robust deep learning model for multi-state emotion classification using electroencephalography (EEG) data.
• To enhance emotion recognition accuracy by integrating brain network analysis with deep neural networks.

Main Methods:

• EEG signals were transformed into five frequency bands using wavelet transform.
• Brain networks were constructed using inter-channel correlation coefficients.
• A dual-pathway deep residual neural network with attention mechanisms processed both network and temporal features.

Main Results:

• The proposed model achieved an average accuracy of 94.57% on a custom dataset.
• Performance on public SEED and SEED-IV datasets reached 94.55% and 78.91%, respectively.
• The model demonstrated superior performance in classifying multiple emotional states.

Conclusions:

• The combined approach of brain network analysis and deep learning effectively recognizes human emotional states.
• The proposed dual-pathway network architecture with attention mechanisms significantly improves emotion recognition accuracy.
• This model shows great potential for real-world emotion recognition applications.