Performance Analysis of Emotion Classification Using Multimodal Fusion Technique

As the central processing unit of the human body, the human brain is in charge of several activities, including cognition, perception, emotion, attention, action, and memory. Emotions have a significant impact on human well-being in their life. Methodologies for accessing emotions of human could be essential for good user-machine interactions. Comprehending BCI (Brain-Computer Interface) strategies for identifying emotions can also help people connect with the world more naturally. Many approaches for identifying human emotions have been developed using signals of EEG for classifying happy, neutral, sad, and angry emotions, discovered to be effective. The emotions are elicited by various methods, including displaying participants visuals of happy and sad facial expressions, listening to emotionally linked music, visuals, and, sometimes, both of these. In this research, a multi-model fusion approach for emotion classification utilizing BCI and EEG data with various classifiers was proposed. The 10-20 electrode setup was used to gather the EEG data. The emotions were classified using the sentimental analysis technique based on user ratings. Simultaneously, Natural Language Processing (NLP) is implemented for increasing accuracy. This analysis classified the assessment parameters as happy, neutral, sad, and angry emotions. Based on these emotions, the proposed model’s performance was assessed in terms of accuracy and overall accuracy. The proposed model has a 93.33 percent overall accuracy and increased performance in all emotions identified.