Emotion recognition in EEG Signals: Deep and machine learning approaches, challenges, and future directions

A crucial part of brain-computer interfaces is the use of electroencephalogram (EEG) signals for human emotion identification, which analyzes patterns of brain activity to determine the emotional state. This field of study is becoming increasingly important for developing advanced applications that enhance brain machine interaction and improve brain health assessment systems. However, EEG signal analysis faces significant challenges due to their subject-specific nature, high noise levels, and the scarcity of high-quality labeled data, which collectively limit model generalizability and complicate signal analysis. Traditional approaches have employed handcrafted features with Support Vector Machines (SVM), K-Nearest Neighbors (KNN), and Random Forests (RF) for EEG feature extraction and classification. Recent advances in deep learning, particularly Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), enable automatic feature learning from raw data to extract temporal, spatial, and spectral properties. The study employs a literature review approach and the analysis of the popular datasets (e.g., DEAP, SEED, AMIGOS). Despite technological advances, the fundamental challenges of noisy subject variability, and limited labeled data persist, requiring future research to focus on improving model robustness, scalability, and interpretability while addressing current limitations.