which utilize unsupervised learning to enhance feature representation before applying supervised learning for emotion classification[5]. This approach not only improves model performance but also ...