<![CDATA[Emotion recognition from electroencephalography (EEG) signals is crucial for human-computer interaction and diagnosing emotional disorders. This study evaluates the impact of feature extraction methods on the performance of XGBoost in classifying emotions in game players using EEG data. It compares the efficacy of Discrete Wavelet Transform (DWT) and Stationary Wavelet Transform (SWT) combined with XGBoost, aiming to identify the most effective feature extraction method for improving emotion classification accuracy. Using the GAMEEMO dataset, which includes preprocessed EEG signals from game players, three scenarios were analyzed: XGBoost without feature extraction, XGBoost with DWT, and XGBoost with SWT. The results demonstrate that DWT significantly enhances classification performance, achieving higher accuracy, precision, and recall compared to SWT and no feature extraction. DWT's ability to capture rapid frequency changes in EEG signals is a key factor in its superior performance. Future work should focus on refining data preprocessing techniques, exploring additional feature extraction methods, and optimizing XGBoost hyperparameters to further enhance emotion recognition accuracy. This research provides valuable insights into the comparative effectiveness of different wavelet transform methods for EEG-based emotion classification, emphasizing the potential of DWT for improved performance]]>
Copyrights © 2024
