基于VAE-GWO-LightGBM的信用卡欺诈检测方法  

Credit card fraud detection method based on VAE-GWA-LightGBM

作　　者：赵峰[1] 李妞妞 ZHAO Feng;LI Niu-niu(School of Management Science&Engineering,Anhui University of Technology,Ma anshan 243032,China)

机构地区：[1]安徽工业大学管理科学与工程学院,安徽马鞍山243032

出　　处：《东北师大学报(自然科学版)》2023年第4期77-84,共8页Journal of Northeast Normal University(Natural Science Edition)

基　　金：国家自然科学基金资助项目(71872002);安徽省高校人文社会科学研究重点项目(SK2019A0072).

摘　　要：针对信用卡欺诈检测中样本数据规模大、计算复杂程度高、数据分布极度不平衡等问题,提出一种结合变分自编码器(VAE)、灰狼算法(GWO)和轻量级梯度提升机(LightGBM)的信用卡欺诈检测(VAE-GWO-LightGBM)方法.对原始数据进行预处理,考虑到数据的不平衡性,采用VAE处理训练样本的数据不平衡问题.在扩充后的平衡数据集上,利用LightGBM作为分类器,并通过智能优化算法(GWO算法)对分类器参数进行寻优,进而获得最优分类器,提高欺诈检测性能.最后在信用卡欺诈数据和其他不平衡数据集上进行对比实验验证.结果表明,基于VAE-GWO-LightGBM的融合模型从整体上大大提高了信用卡欺诈检测的识别率.In order to solve the problem that traditional machine learning methods have low minority class detection rate when dealing with unbalanced and massive abnormal data,a fraud detection method combining variational autoencoder(VAE),Gray Wolf algorithm(GWO)and lightweight gradient hoist(LightGBM)was proposed(VAR-GWA-LightgBM).After preprocessing the original network data and considering the imbalance of the data,VAE was used to expand the data set with minority samples.On the expanded balanced dataset,LightGBM was used as a classifier,and the intelligent optimization algorithm GWO algorithm was used to optimize the parameters of the classifier,and then the optimal classifier was obtained.Finally,the feasibility and effectiveness of the method are verified by simulation and comparison experiments on credit card fraud data and other imbalanced data.

关 键 词：信用卡欺诈 变分自编码器 灰狼算法 轻量级梯度提升机 参数优化 不平衡数据分类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]