Designing a Bankruptcy Prediction System using Function-Link Cerebellar Model Neural Network

The purpose of this article is to design a bankruptcy prediction system by application of Function-Link Cerebellar Model Neural Network (FL-CMNN) as the classifier. FL-CMNN is designed by integrating a standard Cerebellar Model Articulation Controller (CMAC) based neural network with a Function-Link network (FLN) which is used to expand the input space of the neural network architecture. The Function-Link network augments the Cerebellar Model Neural Network by generalizing the architecture and broadening the diversity of its application. Additionally, the FLN provides good function approximation and therefore improving its performance for prediction and classification problems. The performance of the bankruptcy prediction models of the Function-Link Cerebellar Model Neural Network and the classic CMAC are compared using established variables to predict financial distress. The data was derived from financial information published in the Taiwan Economic Journal and the performance of the model is illustrated.