CAJ | 학술논문

针对国内外金融领域可疑交易的低检测率问题,通过对RBF(Radial Basis Function)神经网络技术的分析与研究,提出了一种基于APC-Ⅲ聚类算法和RLS(Recursive Least Square)算法的面向反洗钱的RBF神经网络模型并加以实现.APC-Ⅲ聚类算法用于确定RBF神经网络隐含层的中心向量,RLS算法用来调整隐舍层与输出层之间的连接权值.RBF神经网络与支持向量机(SVM)和孤立点检测相比,有更高的检测率和较低的误检率,因此,提出的模型具有重要的理论和实用价值.
침대국내외금융영역가의교역적저검측솔문제,통과대RBF(Radial Basis Function)신경망락기술적분석여연구,제출료일충기우APC-Ⅲ취류산법화RLS(Recursive Least Square)산법적면향반세전적RBF신경망락모형병가이실현.APC-Ⅲ취류산법용우학정RBF신경망락은함층적중심향량,RLS산법용래조정은사층여수출층지간적련접권치.RBF신경망락여지지향량궤(SVM)화고립점검측상비,유경고적검측솔화교저적오검솔,인차,제출적모형구유중요적이론화실용개치.
Aiming at the low detection rate of suspicious transaction at home and abroad in financial field,and with the analysis of Radial Basis Function(RBF) Neural Network,a RBF Neural Network model based on APC-Ⅲ clustering algorithm and Recursive Least Square(RLS) algorithm for anti-money laundering is proposed.APC-Ⅲ clustering algorithm is used for determining the pa-rameters of RBF in hidden layer,and RLS algorithm is adopted to update weights of connections between hidden layer and out-put layer.The proposed method is compared against Support Vector Machine(SVM) and outlier detection methods,which show that the proposed method has the highest detection rate and the lowest false positive rate.Thus the model is proved to have both the-oretical and practical value.