Advanced Hybrid Model: Multimodal Emotion Recognition with Improved EEG Feature Fusion

Emotion is a common occurrence in daily life as a subjectively psychological reaction to environmental stimuli. Emotion recognition via EEG data is quickly becoming a multidisciplinary research subject due to advances in artificial intelligence and brain science. This study compares several machine learning techniques, including Support Vector Machine (SVM), K-nearest neighbour, Linear Discriminant Analysis, Logistic Regression, and Decision Trees. Existing works achieves enhancement on separate/single analysis based models. Each of these models is tested both with and without principal component analysis (PCA) for reducing the dimensionality. This work achieves sustainable computing on all models. Moreover, the epoch data from EEG sensor channels is analysed. This work analysed that in the 30th to 45th segmentation interval, PCA with SVM performed the best, yielding an F1-score of 84.73% with 98.01% recall. This paper summarises current advances and identifies important directions for future study, making it a useful resource for scholars and practitioners in the domains of machine learning and EEG analysis.