Quaternary classification of emotions based on electroencephalogram signals using hybrid deep learning model

• Published in: Journal of ambient intelligence and humanized computing 2023-03, Vol.14 (3), p.2429-2441
• Main Authors: Singh, Khushboo, Ahirwal, Mitul Kumar, Pandey, Manish
• Format: Article
• Language: English
• Subjects: Accuracy; Arousal; Artificial Intelligence; Artificial neural networks; Brain research; Classification; Computational Intelligence; Datasets; Deep learning; Electroencephalography; Emotion recognition; Emotions; Engineering; Machine learning; Neural networks; Original Research; Physiology; Robotics and Automation; Support vector machines; User Interfaces and Human Computer Interaction; Wavelet transforms
• ISSN: 1868-5137; 1868-5145
• DOI: 10.1007/s12652-022-04495-4

Recognizing emotions from electroencephalography (EEG) signals is a trustworthy and reliable method to monitor the mental health of patients and the enthusiasm of individual behavioral feelings and polarity. Recently, the researcher focused on classifying EEG emotions from multimodal DEAP data set into binary (high and low) and ternary (low, medium and high) classes based on valence and arousal scale. However, for deep and intrinsic emotion recognition, multiclass classification is preferred. But non-linearity and non-stationary nature of EEG make it more challenging. Therefore, this paper proposes a hybrid deep learning model of one-dimensional Convolutional Neural Network (1DCNN) and Bidirectional Long Short-Term Memory (BI-LSTM) for multiclass emotion classification. The hybrid model linearly distributed desired class labels over two-dimensional emotion space (valence and arousal) as happy, relaxed, anger, and sad. The simulation results of the proposed model for binary, ternary and quaternary classification of emotion acquire 91.31%, 89.32% and 88.19% accuracy, respectively. Subject wise analysis of results has also been performed. The novelty of this work is the quaternary classification of emotions.