ANN-BASED CREDIT RISK IDENTIFICAION AND CONTROL FOR COMMERCIAL BANKS
To provide rational, intelligent and real time decision supports to credit risk management for commercial banks, an ANN-based credit risk identification and control method was proposed. A credit risk measurement indicator system was established incorporating both internal and external related factors of debtor firms. And credit risk was identified based on the online learning of an ANN model. To meet the online learning requirement, an improved BP training algorithm with adaptive learning rate and momentum was proposed for speed enhancement. The ANN-based model proposed is suitable for Chinese commercial banks, which only have limited and incomplete historical data due to lagged credit risk management. The model can represent the experience, knowledge and intuitiveness of the experts. And with data accumulation over time, the identification results can be improved through online learning of the ANN model, ensuring objectiveness, rationality and timeliness. An example is given to illustrate the method.
Artificial Neural Network (ANN) Back Propagation (BP) Credit Risk Commercial Bank
XIN-YUE HU YONG-LI TANG
School of Management, Jinan University, Guangzhou 510632, Guangdong P.R.China
国际会议
2006 International Conference on Machine Learning and Cybernetics(IEEE第五届机器学习与控制论坛)
大连
英文
3110-3114
2006-08-13（万方平台首次上网日期，不代表论文的发表时间）