Orthogonal Least Square Based Support Vector Machine for the Classification of Infant Cry with Asphyxia

This paper describes the classification of asphyxiated infant cry using orthogonal least square (OLS) based Support vector machine (SVM). The features of the cry signal were extracted using mel frequency cepstral coefficient analysis and significant features were selected using OLS. SVM with linear and RBF kernels were used to classify the asphyxiated infant cry signals. Classification accuracy and support vector number were computed to examine the performance of the OLS based SVM. The highest classification accuracy (93.16%) could be achieved using RBF kernel, however, with large support vector number.