A Radial Basis Function Network Oriented for Infant Cry Classification

Several investigations around the world have been postulated that the infant cry can be utilized to asses the infant’s status and the use of Artificial Neural Networks (ANN) has been one of the recent alternatives to classify cry signals [4,9]. A Radial Basis Function (RBF) network is implemented for infant cry classification in order to find out relevant aspects concerned with the presence of some CNS diseases. The cry episodes were obtained from 35 neonates, nineteen healthy cases and sixteen sick neonates with hypoxia-based CNS diseases. First, an intelligent searching algorithm combined with a fast nonlinear classification procedure is implemented, establishing the cry parameters which better matches the physiological status previously defined for the six control groups used as input data. Finally the optimal acoustic parameter set is chosen in order to implement a new nonlinear classifier based on a radial basis function network, an ANN-based procedure which classifies the cry units into a 2 categories, normal-or abnormal case. All the experiments were based on the physioacoustic model for cry production and the Golub’s muscle control model [6,8].