Research on Emotion Recognition of EEG Signal Based on Convolutional Neural Networks and High-Order Cross-Analysis

Area of Science:

• Neuroscience
• Artificial Intelligence
• Biomedical Engineering

Background:

• Emotion recognition is crucial for human-computer interaction.
• Electroencephalogram (EEG)-based methods are effective for real-time emotion detection.
• Convolutional Neural Networks (CNNs) show promise in analyzing complex physiological signals.

Purpose of the Study:

• To classify EEG signals into distinct emotional states using CNNs.
• To investigate the efficacy of single-channel feature extraction for emotion recognition.
• To evaluate CNN performance on the Database for Emotion Analysis using physiological signals (DEAP) dataset.

Main Methods:

• Utilized a seven-layer CNN for classifying 32-channel EEG data.
• Extracted high-order cross-feature samples from single EEG channels.
• Employed the DEAP dataset for training and validation.

Main Results:

• Achieved an average accuracy of 65% for classifying four emotional states and 58.62% for three states using 32-channel EEG.
• Single-channel classification yielded an average accuracy of 43.5% for four emotional states.
• The F4 channel demonstrated the highest classification accuracy at 44.25%; even-numbered channels outperformed odd-numbered ones.

Conclusions:

• CNNs can effectively classify emotions from EEG signals.
• High-order cross-feature extraction from single channels shows potential, though multi-channel approaches yield higher accuracy.
• The findings support the development of real-time EEG-based emotion recognition systems.