Support Vector Machine and Neural Network Algorithm Approach to Classifying Facial Expression Recognition

One of the obstacles to detecting facial emotions is the lack of analysis in the process of emotional expression on human faces based on a photo camera. Identifying features and implementing different feature combinations can improve accuracy, in most cases. Trial detection through the use of feature vectors with higher dimensions that contain facial emotions can provide a lot of information on the accuracy of the data. The purpose of this study is to determine the highest accuracy results from the comparison of Support Vector Machine and Neural Network Algorithms. It can be seen that the Support Vector Machine has the highest accuracy value, which is 87%, while the Neural Network algorithm only has an accuracy of 85%. The results of the ROC Curve show that the Support Vector Machine achieves the best AUC value, namely 0.97. Comparison between Neural Network algorithm and Support Vector Machine for prediction of emotions using data types with many variations of emotions including anger, sadness, happy and surprise. The Support Vector Machine method is an accurate algorithm and this method is also very dominant over other methods. Based on the Accuracy, AUC, and T-test method, this method falls into the best classification.