作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2008, Vol. 34 ›› Issue (15): 205-207,. doi: 10.3969/j.issn.1000-3428.2008.15.074

• 人工智能及识别技术 • 上一篇    下一篇

基于置信度和神经网络的信用卡异常检测

郭 涛,李贵洋,袁 丁   

  1. (四川师范大学计算机科学学院,成都 610066)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2008-08-05 发布日期:2008-08-05

Abnormity Detection of Credit Card Based on Confidence and Neural Network

GUO Tao, LI Gui-yang, YUAN Ding   

  1. (College of Computer Science and Technology, Sichuan Normal University, Chengdu 610066)
  • Received:1900-01-01 Revised:1900-01-01 Online:2008-08-05 Published:2008-08-05

摘要: 针对信用卡使用过程中存在的异常消费行为,提出一种新的基于置信度和神经网络的信用卡异常检测方法以及采用ROC分析技术确定系统检测阈值方法,实现了对消费行为特征属性数据的置信度计算,利用BP神经网络建立了信用卡消费行为异常检测模型。实验结果表明,该检测模型不依赖于单个持卡用户,ROC分析技术的引入确保了检测的准确性和有效性,系统的实现对信用卡异常检测有较好的实用性和适应性。

关键词: 信用卡异常, 神经网络, ROC分析技术, 消费模式

Abstract: A new detection method with confidence-based neural network and ROC analysis technology for determining threshold is proposed in this paper. With the scheme, the properties of spending pattern are converted to confidences. The abnormal detection model for credit card is established with algorithm of BP neural network. Test results show that the detection model is not dependent on a single card user and the introduction of ROC analysis technology ensures the accuracy and effectiveness of the abnormal detection. The model possesses preferable practicability and self-adaptability

Key words: credit card abnormity, neural network, ROC analysis technology, spending pattern

中图分类号: