Enhancing EEG-Based Emotion Detection with Hybrid Models: Insights from DEAP Dataset Applications.

Area of Science:

• Neuroscience
• Computer Science
• Artificial Intelligence

Background:

• Emotion detection from electroencephalogram (EEG) signals is crucial for mental health and human-computer interaction.
• Existing methods struggle with accuracy, interpretability, and real-time processing.

Purpose of the Study:

• To evaluate machine learning and deep learning techniques for emotion recognition from EEG signals.
• To develop a computationally efficient and interpretable real-time emotion detection system.

Main Methods:

• Experimentation with various algorithms including KNN, SVM, DT, RF, BiLSTM, GRUs, CNNs, autoencoders, and transformers on the DEAP dataset.
• Development of a hybrid deep learning approach.
• Application of SHapley Additive exPlanations (SHAP) for model interpretability.

Main Results:

• A hybrid approach achieved peak accuracy of 85-95% in emotion recognition from EEG signals.
• The developed system demonstrates advantages in computational efficiency and real-time applicability.
• SHAP analysis provided insights into feature contributions, enhancing model interpretability.

Conclusions:

• Hybrid deep learning models effectively improve accuracy, interpretability, and real-time processing for EEG-based emotion detection.
• The developed real-time system is feasible for practical deployment in neurofeedback, mental health monitoring, and affective computing.
• Future work includes dataset expansion and testing on diverse populations for broader applications.