CAJ | 학술논문

随着新巴赛尔协议的推出，信用风险测量问题越来越受到重视。在西方发达国家，商业银行的信用风险管理测量技术已比较成熟，继传统的比例分析、主观分析之后，统计方法得到广泛的应用，如判别分析和logit回归分析等。自从20世纪80年代末期以来，人工智能技术如神经网络、专家系统等也被应用于商业银行信用风险测量中。目前，该领域应用最多的就是BP神经网络，但其固有的一些缺点，如易陷入局部极小点，会影响预测效果。但可以利用改进的遗传算法对BP神经网络进行优化，实验证明效果非常好。
수착신파새이협의적추출，신용풍험측량문제월래월수도중시。재서방발체국가，상업은행적신용풍험관리측량기술이비교성숙，계전통적비례분석、주관분석지후，통계방법득도엄범적응용，여판별분석화logit회귀분석등。자종20세기80년대말기이래，인공지능기술여신경망락、전가계통등야피응용우상업은행신용풍험측량중。목전，해영역응용최다적취시BP신경망락，단기고유적일사결점，여역함입국부겁소점，회영향예측효과。단가이이용개진적유전산법대BP신경망락진행우화，실험증명효과비상호。
Credit risk measurement becomes more important with the new Basel Capital Accord＇sh publicizing. In developed countries, the technologies of credit risk measurement are mature. After traditional scale analysis and subjective analysis, statistic methods are used widely, for example discriminant analysis and Logit regression analysis. Artificial intelligence technologies as Neural Network and expert systems are also used in the field from 1980g. Now Back - Propagation Neural Network is used the most frequently in this field. But it has some shortcomings as likelihood of getting in local point. So effect will be impacted. This paper optimizes Back - Propagation Neural Network with improved Genetic Algorithm, and the result is good.