Electroencephalography-based emotion recognition with empirical mode decomposition and ensemble machine learning methods

Abstract

Emotion recognition stands as one of the most challenging tasks in pattern recognition, machine learning, and artificial intelligence. Incorporating emotion recognition in brain–computer interfaces (BCIs) is a recent trend. In fact, this phenomenon makes BCI systems more sensitive, flexible, and supportive of users’ emotional and cognitive demands. Emotion recognition leverages voices, images, and electroencephalography (EEG) signals for an automated identification of emotions, proving particularly valuable in diverse sectors. In today’s digital era, providing accurate insights into emotion recognition is crucial. Given the complexity of emotional activity, the application of advanced technologies and the utilization of signal processing and machine learning methodologies are essential for an effective analysis. Despite ongoing efforts to recognize emotional activities over the past decade, fundamental issues remain that need to be addressed to fully harness technology in understanding emotional states. This study explores recent advancements in signal processing and machine learning algorithms tailored for detecting emotional activity. It also discusses the challenges and critical considerations inherent in emotion recognition. Additionally, the chapter introduces several open concepts aimed at guiding future research efforts in addressing these challenges. Finally, specific examples of emotion recognition using EEG signals are presented, showcasing various AI and signal processing techniques employed in this domain.