Deep Learning Techniques for EEG-Based Emotion Recognition: A Systematic Review

This is the data associated with the systematic review of EEG-based emotion recognition using deep learning architectures. The review is conducted on studies published between 2020 and March 2025 and follows the PRISMA guidelines. The study categorizes unimodal EEG-based emotion recognition research into supervised, unsupervised, and hybrid approaches. In addition, the commonly used public datasets in these studies and the preprocessing steps are described. Studies were identified through five major databases, and their methodologies, evaluation metrics, and outcomes were systematically analyzed. The main challenges in using EEG signals for emotion recognition are highlighted, and future research directions are proposed to enhance generalizability, interpretability, and data efficiency.