EEG-Based Emotion Recognition Using a 2D CNN with Different Kernels.

Area of Science:

• Neuroscience
• Computer Science
• Artificial Intelligence

Background:

• Emotion recognition is crucial in healthcare and Human-Computer Interaction (HCI).
• Electroencephalogram (EEG)-based methods are widely used due to their correlation with emotions and ability to influence external expressions.
• Existing machine learning approaches for EEG emotion detection face challenges including manual feature extraction and limited model generalization.

Purpose of the Study:

• To propose a novel deep neural network for EEG-based emotion classification.
• To overcome the limitations of traditional machine learning methods in EEG emotion recognition.
• To improve the accuracy and universality of emotion recognition models.

Main Methods:

• Development of a novel 2D Convolutional Neural Network (CNN).
• Utilizing two convolutional kernels of different sizes for feature extraction in temporal and spatial dimensions.
• Employing the public DEAP dataset for experimental validation.

Main Results:

• Achieved high accuracies of 99.99% for arousal and 99.98% for valence in binary classification.
• Demonstrated the effectiveness of the proposed 2D CNN in extracting emotion-related features.
• Validated the model's performance on a public EEG emotion dataset.

Conclusions:

• The proposed deep neural network effectively addresses limitations of previous EEG emotion recognition studies.
• The model shows significant potential for advancing research and applications in emotion recognition.
• High accuracy suggests a promising direction for developing more generalized and accurate emotion detection systems.