Classification Credit Dataset Using Particle Swarm Optimization and Probabilistic Neural Network Models Based on the Dynamic Decay Learning Algorithm

Abstract

This paper describes a credit risk evaluation system that uses supervised probabilistic neural network (PNN) models based on the Dynamic Decay learning algorithm (DDA). The PNN-DDA has two parameters called positive and negative threshold. This learning algorithm trains very quickly. Thus it makes sense that we use a meta-heuristic algorithm such as particle swarm optimization to optimize these parameters. When using the meta-heuristic algorithm such PSO, the tuning process of parameters is implemented wisely. Thus in this paper we also obtained optimum threshold. Two credit datasets in UCI database are selected as the experimental data to demonstrate the accuracy of the proposed model. The result shows that this new hybrid algorithm outperforms the most common used algorithm such as multi-layer neural network.