Comparative Analysis of Machine Learning Approaches for Emotion Recognition Using EEG and ECG Signals

Emotions significantly influence human behaviour and decision-making, particularly in a digital era dominated by human-computer interactions (HCIs). Emotion can be expressed in various forms, including facial expressions, textual descriptions, and physiological responses. The main objective of this study is to comparatively analyze the performance of various machine learning (ML) classifiers to accurately recognize human emotional states using electroencephalogram (EEG) and electrocardiogram (ECG) signals. This study uses the DREAMER dataset and classifies emotional state in four different ways according to valence, arousal, and dominance (VAD) values – binary emotions, positive-neutral-negative (PNN) emotions, two-dimensional valence-arousal emotional space, and three-dimensional VAD emotional space. An ML pipeline has been developed to detect human emotions with EEG and ECG signals. Without removing outliers and balancing the dataset, the classifier that achieved the best performance was the ensemble classifier (SVM + random forest). If emotion is defined as a binary state, our experimental results show that both the SVM and the ensemble classifiers strike a good performance with approximately 80% accuracy; however, they perform poorly with the non-binary emotional models. The multinomial logistic regression (MLR) classifier and the random forest (RF) classifier consistently achieve a good performance for both the binary and the non-binary emotion models with 80% - 90% accuracy, its accuracy is higher than the accuracy in the original DREAMER experimental results. Our study experimentally confirmed this obvious finding.