Advanced Emotion Recognition: A Heuristic Approach Applied to EEG Signals Using Machine Learning

Abstract

Emotion analysis through electroencephalographic (EEG) signals has become a prominent research focus due to its applications in fields such as marketing, education, and mental health. Despite numerous methods available for emotion recognition, there remains a lack of robust metrics to validate the accuracy of these analyses against the actual emotional states. This study presents a novel heuristic approach for emotion analysis using EEG signals, employing an advanced algorithm that enhances the normalization of Valence and Arousal values through the Emotiv Epoc+ device. The algorithm not only refines these critical variables but also incorporates context-specific adjustments within an improved database schema, allowing for a more adaptive and precise evaluation of emotional states. Comparisons were made with the Self-Assessment Manikin (SAM) test, a validated tool in psychology, to verify the physiological responses recorded by the EEG signals. Initial findings demonstrated an accuracy of 76.47%, which increased to 79.45% after implementing the proposed enhancements, validated using the DBSCAN clustering algorithm. This study effectively demonstrates the algorithm’s capacity to classify emotional states in a sample of 15 participants aged 16 to 25 years, highlighting the potential of this heuristic approach in enhancing the reliability and applicability of EEG-based emotion recognition. The proposed methodology not only improves the accuracy of emotion detection but also establishes a foundation for integrating specific contextual factors into EEG analysis, thereby expanding its application in brain-computer interfaces, mental health monitoring, and other advanced research areas. These findings underscore the value of combining physiological data with validated psychological assessments, offering a significant advancement in the field of emotion recognition.