| Optimizing Hybrid Classification Using 1D-CNN and LSTM for EEG-Based Emotion Recognition |
| Paper ID : 1041-ICEEM2025 (R1) |
| Authors |
|
Ahmed Mohamed Fouad Galal *, Mahmoua Ahmed Attia Ali, Heba Ali El-KHobby Tanta University Faculty of Engineering |
| Abstract |
| Abstract—This research aims to enhance the ability of computers to accurately classify emotions through the analysis of brain signals. The emotional state is a complex mixture of feelings, and any change in an individual's emotions can significantly impact their life. We will provide the computer with healthy and valid emotional data extracted from the DEAP dataset, which includes information from individuals who do not suffer from psychological disorders. This will enable the computer to conduct accurate automatic emotional analysis of received brain signals, allowing for precise identification of an individual's emotional state. Such capabilities will enhance diagnostic potential and facilitate restoring individuals to their normal state before any deterioration occurs. This study focuses on feature extraction from brain signals in a straightforward manner, utilizing parallel analysis, which has shown to yield better results than sequential analysis. We selected two models, DCNN and LSTM, based on previous research demonstrating their superior accuracy compared to other models. The results achieved a precision rate of 99.8%, indicating the effectiveness and capability of our method in accurately classifying emotional states. This research paper will contribute to various fields by enhancing the understanding and knowledge of individuals' psychological and health conditions. |
| Keywords |
| EEG, emotion recognition, CNN, LSTM, majority voting. |
| Status: Accepted |