Hybrid CNN-SVM Classifier for Emotion Recognition Using Brain Waves

The field of information fusion and deep learning (DL) has advanced significantly in recent years, enabling computers and other devices to perceive, recognize, and evaluate emotions. In addition to distinct physiological rhythms and alterations in behavior, mental health, and physical attributes, emotions constitute a mental state. The first three techniques might not be effective because people occasionally accidentally or purposely hide their true feelings. We call this social masking phenomena. By using physiological clues, emotion detection could become more objective and reliable. Important features of emotional states can be inferred from electroencephalography (EEG) signals since they react to changes in emotional states faster and more sensitively than peripheral neurophysiological signals. Automatic electroencephalography (EEG) emotion identification is one challenging component of human-computer interaction (HCI). Inspired by the powerful feature learning capabilities of recently emerging deep learning algorithms, several advanced deep learning models have been used more often to obtain high-level feature representations for EEG emotion recognition. In order to classify EEG emotions more quickly and accurately, this research introduces a hybrid model based on convolutional neural networks (CNN) and support vector machines (SVM). Experimental results show that the suggested strategy outperforms previous approaches in modeling training with a smaller loss value and more generalization. The categorization accuracy as a whole was 98.361%. This percentage is higher when compared to other learning strategies. Furthermore, compared to other models, the proposed model takes less time and much less complex.