Neuro-genetic approach for bankruptcy prediction modeling

Abstract

Artificial neural network (ANN) modeling has become the dominant modeling paradigm for bankruptcy prediction. To further improve the neural network's prediction capability, the integration of the ANN models and the hybridization of ANN with relevant paradigms such as evolutionary computing has been demanded. This paper first attempted to apply neuro-genetic approach to bankruptcy prediction problem for finding optimal weights and confirmed that the approach can be a good methodology though it currently could not outperform the backpropagation learning algorithm. The result of this paper shows a possibility of neuro-genetic approach to bankruptcy prediction problem since the simple neuro-genetic approach produced a meaningful performance. © Springer-Verlag 2004.