Radial Basis Function Neural Network Optimizating And Its Application In Corporate Financial Distress Prediction

In recent years, artificial intelligence theory and data mining technologies havemade tremendous success in pattern recognition and prediction along with theirdevelopment and maturity. Applications in the financial area with data miningtechnologies have been researched widely. Radial-basis function neural network(RBFN) based on regularization theory is a feedforward network with three layers. Ithas developed rapidly in these years because of the merits of strong generalizationability, fast convergence and without local minimum.Compared with traditional neural network models, radial basis function neuralnetwork employed the parameters training approach by stages, thus avoiding thecomplicated calculation process such as the BP network, eventually improving thetraining efficiency of the network models, radial basis function neural networks mayperform poorly when given too much input to it, because of the extension of the centerand the linear parameter training process, thus reducing the efficiency of networktraining. In this study, principle component analysis based radial basis functionnetwork was proposed, in which principle component analysis is used firstly to initialtraining set for feature selection, thereby enhanced the training and predictingefficiency of the model through reducing the number of input nodes.A detailed review and study indicated, here radial basis function neural networkwhich was optimized by principle component analysis was proposed in the field ofcorporate financial distress prediction, the middle-sized company’s financial indicatorof actual data was applied in this article. We make experiment through the softwarematlab to test the effectiveness of the proposed improving method. The final resultshave corroborated that radial basis function neural network which was optimized byprinciple component analysis do good to improve the accuracy of financial statusclassification prediction. Thus in theory and practice of financial distress predictionmodel for our development and further improve the provision of a strong reference.