ENT-DASM and ENT-RASM: Novel Energy-Entropy Asymmetry Features for Enhanced EEG-Based Emotion Classification

Abstract

Electroencephalography (EEG) based emotion recognition is crucial for advancing brain-computer interfaces and affective computing. However, existing feature extraction methods often fail to capture the complex emotional information in EEG signals. This study introduces novel energy-entropy based features, ENT-DASM and ENT-RASM, which combine differential and rational asymmetry measures with energy and entropy calculations. We evaluate these features on the DEAP dataset using various machine learning classifiers across different brain regions and frequency bands. Our method achieves a maximum accuracy of 83.24% for the High Arousal Low Valence (HALV) emotional state using Support Vector Machines, outperforming existing approaches by 1-2%. The proposed features demonstrate consistent performance across all frequency bands, particularly excelling in beta and gamma ranges. This study contributes to the field by providing more robust and informative features for EEG-based emotion classification, potentially improving applications in mental health monitoring and adaptive user interfaces.