Comparative studies in methods of feature recognition with machine learning for affective computing: A survey

A survey study about the various methods of feature recognition with machine learning for affective computing is examined. In order to explore the methods of feature recognition with machine learning methods, Sequential Floating Forward Selection (SFFS), Minimum Redundancy – Maximum Relevance (mRMR), Information Gain(IG), and Fisher projection (FP) are discussed. As the machine learning methods, k-Nearest Neighbor (kNN), Support Vector Machine (SVM), and Multilayer Perceptron (MLP) are described. Then, the various feature recognition methods with machine learning methods are compared by the statistical analyses using the classification accuracy performance with applying the discrete emotion data.