A Review of Deep Learning Techniques for EEG-Based Emotion Recognition: Models, Methods, and Datasets.

Area of Science:

• Affective computing
• Neuroscience
• Artificial Intelligence

Background:

• Emotion Recognition (ER) using Electroencephalography (EEG) is crucial for understanding brain activity.
• Deep Learning (DL) has significantly improved ER accuracy and efficiency with EEG data.

Purpose of the Study:

• To systematically review Deep Learning (DL) approaches for EEG-based Emotion Recognition (ER).
• To analyze public datasets used in EEG ER studies.
• To guide future research towards more interpretable, generalizable, and data-efficient EEG ER systems.

Main Methods:

• Systematic literature review following PRISMA guidelines.
• Comprehensive search across five major databases (2020-2025).
• Inclusion of studies utilizing EEG signals and DL architectures for ER.

Main Results:

• 233 articles were included after eligibility screening.
• Analysis of public datasets regarding stimulation and emotional representation.
• Identification of trends and gaps in DL-based EEG ER research.

Conclusions:

• DL shows significant promise for EEG-based ER.
• Further research is needed to improve interpretability, generalizability, and data efficiency.
• This review provides a roadmap for developing reliable and scalable EEG-driven ER systems.