基于多进化神经网络的信用评估模型研究

客户信用评佑对于银行的经营管理有着重要的意义，为此提出了一种基于多进化神经网络的信用评估模型(MNN-CREDIT)。该模型基于客户信货数据，利用基于聚类的小生境遗传算法并行地训练出多个精度高、差异性大的三层前馈神经网络，然后将待识别的客户数据分别输入，最后根据动态投票法集成最终信用预测结果。利用德国信用数据库真实数据集进行了实证分析，结果表明，基于多进化神经网络的信用评估模型具有较高的预测精度。

Credit Evaluation Model Based on Multiple Evolutionary Neural Networks

Credit evaluation plays an important role in the banking management. A novel credit evaluation model based on multiple evolutionary neural networks, named MNN-CREDIT, was presented. The MNN-CREDIT model establishes classifiers by a group of three-layer feed-forward neural networks with high accuracy and good diversity. The neural networks arc trained by niche genetic algorithm based on clustering. The credit evaluation result of the identifying client can first be evaluated by each neural network, and the final credit classification result is obtained according to the dynamic voting rule. Empirical analysis on Germany credit database was given. hhc results show that MNN-CREDIT model has higher prediction precision.