【摘要】 巴塞尔新资本协议中对信用风险评估使用的数据作了明确规定,而我国银行业目前所积累的数据还不能达到协议的要求;立足于我国银行业的现状,在新协议框架范围内探索可行的信用评估方法。对原有神经网络算法加以改进,提出自优化神经网络方法,该方法能较好地适应时变数据和自动优化神经网络评估模型,用真实的客户信用数据试验也表明该方法比普通神经网络有较高的准确性。更多还原

【Abstract】 The needs of the data for credit assessment has been specified in New Basel Capital Accord, but the data used in Chinese banks cannot meet these needs. According to the present conditions in our banks, practicable methods are studied in the limit of New Basel Capital Accord. A self-optimization neural networks method which could adapt data changing by time and self-optimize assessment model was proposed. The test using real credit data proved that the method has higher veracity than common neural networks.更多还原