The KNN-Smote-LSTM Based Credit Card Fraud Risk Detection Network Model

Financial technology is constantly pushing for a comprehensive upgrade of payment methods.Big data,Internet of Things,cloud computing,artificial intelligence and other technologies continue to pay in the payment field,which has a far-reaching impact on the payment format,providing more security and convenience for our daily lives,and bringing efficiency and value to the business.Improvement.In recent years,with the rapid development of mobile Internet technology,the widespread application of smart terminals and the rapid development of electronic payment services,many banks have begun to support online applications and online lending.Credit cards with consumer credit as their main function have become very common.Financial tools,and through consumer finance such as credit cards,many Internet finance applications,such as ant flower buds,Jingdong white bars,360 borrowings,etc.,are constantly becoming the main medium in the payment field.However,at the same time,fraudulent transactions have also grown at an alarming rate,and fraudulent means have been continuously refurbished,which has greatly disrupted the normal financial order and restricted the long-term healthy development of the financial industry.Studying the consumer financial risk monitoring model is of great significance for improving the financial market system,promoting the healthy development of a virtuous cycle of economy and finance,maintaining the sustained and stable development of the national economy,and ensuring national financial security.As the core technology of artificial intelligence,deep learning has achieved remarkable results in the fields of image,voice and natural language processing.This paper takes credit card fraud detection as an example to elaborate the problem of credit card fraud detection and unbalanced classification.Based on the existing credit card fraud detection models and methods at home and abroad,this paper studies credit card fraud based on deep learning and oversampling algorithm.Detection model,this paper uses the Long Short Term Memory Networks(LSTM)fraud detection model to classify the transaction data,and integrates the Synthetic Minority Oversampling Technique(Smote)algorithm and K nearest neighbor.(k-Nearest Neighbor,kNN)classification algorithm designed and constructed a credit card fraud detection model based on kNN-Smote-LSTM,which can continuously filter out safely generated samples to improve the performance of the model through kNN discriminant classifier,and overcome the Smote The blindness and limitations of the algorithm in generating new samples.Finally,this paper uses real credit card historical transaction data and compares and verifies the existing classification algorithms and oversampling algorithms through reasonable evaluation criteria,and confirms the feasibility of kNN-Smote-LSTM based credit card fraud detection model.And effectiveness,this model demonstrates superior consumer financial risk detection performance,and provides a theoretical basis and practical reference for financial institutions to apply deep learning technology for consumer financial risk detection.