Facial Expression Recognition: A Machine Learning Approach with SVM, Random Forest, KNN, and Decision Tree Using Grid Search Method

This research explores the application of deep learning techniques for emotion recognition using bone-conducted (BC) speech, utilizing the EmoBone dataset—the first dataset specifically created for this purpose. The study aims to investigate the effectiveness of deep learning ap- proaches in recognizing emotions from BC speech while addressing challenges such as degradation and informa- tion loss in neural networks. By implementing advanced models, the research compares the accuracy of emotion recognition with other relevant techniques, showcasing the advantages of BC speech in terms of noise resilience and privacy. The findings contribute to the develop- ment of more accurate and reliable emotion recognition systems, creating the way for innovative applications in fields such as healthcare, human-computer interaction, and secure communication systems.