EEG-ERnet: Emotion Recognition based on Rhythmic EEG Convolutional Neural Network Model.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Signal Processing

Background:

• Emotion recognition using electroencephalography (EEG) is crucial for advancing brain-computer interfaces (BCIs).
• Deep learning models like CNNs show promise but struggle with distinguishing brain rhythm characteristics and temporal dynamics.
• Subject-independent emotion recognition is necessary due to individual variability in emotional responses.

Purpose of the Study:

• To develop a novel network model for accurate and subject-independent emotion recognition from EEG signals.
• To address the limitations of standard CNNs in capturing distinct brain rhythms and temporal variations.
• To enhance the performance of emotion recognition systems by considering individual differences.

Main Methods:

• Proposed a novel network model utilizing depthwise parallel convolutional neural networks (CNNs).
• Extracted power spectral densities (PSDs) from various brain rhythms and projected them as 2D images.
• Developed the EEG-ERnet (Emotion Recognition Network) to process these rhythmic image representations for emotion classification.

Main Results:

• Demonstrated that emotion-specific rhythms within 5-second intervals effectively support emotion classification.
• Achieved high average classification accuracies on the DEAP dataset: 93.27% for valence, 92.16% for arousal, 90.56% for dominance, and 86.68% for liking.
• Validated the model's effectiveness using 10-fold cross-validation.

Conclusions:

• Provided valuable insights into the rhythmic characteristics of emotional EEG signals.
• The EEG-ERnet model shows significant potential for developing efficient, subject-independent emotion-aware systems.
• The proposed model paves the way for portable and real-world emotion recognition applications.