节点文献
概率神经网络在可疑交易监测中的应用及效率比较
Application of probabilistic neural network to suspicious financial transaction surveillance and its efficiency comparison
【摘要】 通过概率神经网络PNN对金融交易时间序列数据的预测偏移误差分类实现对交易异常与否的分类,并将其与前馈神经网络BP、后馈神经网络Elman、竞争型神经网络LVQ、SOM等4种经典类型的分类效率进行比较,结果发现PNN在相近预测精度前提下在网络结构、运行效率方面都有明显优势,适合金融交易海量数据的监测分析.
【Abstract】 We apply the probabilistic neural network(PNN) to suspicious financial transaction surveillance by deviation of time series data prediction.Comparing with other type neural networks including BP,Elman,LVQ and SOM neural netowrks,we find that PNN has obvious advantages in structural complexity and efficiencies which are suitable for analysis of massive financial data.
【关键词】 概率神经网络;
可疑金融交易;
离群检测;
【Key words】 probabilistic neural network; suspicious financial transaction; outlier detection;
【Key words】 probabilistic neural network; suspicious financial transaction; outlier detection;
【基金】 国家社科基金项目(编号:09BTJ002)
- 【文献出处】 武汉大学学报(工学版) ,Engineering Journal of Wuhan University , 编辑部邮箱 ,2013年02期
- 【分类号】TP183
- 【被引频次】4
- 【下载频次】163