神经网络在信贷资产风险映射中的应用

综合研究了商业银行信贷资产风险因素和风险水平之间映射与其中的经验贝叶斯决策，将神经网络应用于这个不确定性和强非线性的动态映射，应用超线性收敛方法加快了神经网络的收敛速度，神经网络的训练样本函盖除农业和外贸外的大部分行业，具有比较普遍的代表地训练样本的学习精度已达到中国国有商业银行在实际风险管理中的精度要求，应用风险映射神经网络更接近实时分析，使商业银行能够对信贷资产的风险水平进行连续动态分析，并可根据商业银行信贷资产风险发展和变化的规律性实施风险的主要管理和全面管理。

Application of Neural Network in Risk Mapping of Credit Assets

The mapping from the risk factors to the risk level of the credit assets of commercial banks, and the empirical Bayesian decision in the mapping were synthetically studied. The neural network was applied in this uncertain and strong non linear dynamical mapping. The convergence speed is accelerated by the use of super linear converge approach. The training samples cover most industries except agriculture and foreign trade industry, and make common sense. The precision of the training samples and the neural network mapping is high enough according to the operating requirement of the risk management in Chinese state owned commercial banks. The application of the neural network of mapping from the risk factors and the risk level can access the real time analysis. Commercial banks can continually and dynamically analyze the risk level of their credit assets, and can implement the initiative control and overall control on the risk according to the regularity of the risk.