Research On Credit Card Fraud Prediction Model Based On Generative Countermeasure Network

With the gradual development of people's advanced consumption habits,the credit card industry has developed rapidly.At present,credit card consumption has become a very important means of consumption in modern life.With the rapid development of credit card industry,credit card fraud occurs from time to time.Credit card fraud has greatly disturbed the normal financial order and restricted the healthy development of the credit card industry.It is urgent to find an effective method to predict credit card fraud.Because the credit card transaction data is a kind of seriously unbalanced data,the number of normal transaction samples is far more than fraud samples,so how to effectively deal with the unbalanced data and how to establish a prediction model is the key to the success of credit card fraud prediction.Based on the background of credit card fraud prediction,this thesis expounds the classification of unbalanced data and the prediction of credit card fraud in detail.Then,this thesis proposes a credit card fraud prediction model based on generative adversarial network according to the analysis of machine learning algorithm and sampling technology.In this study,some samples of fraud are generated by the generative adversarial network,and then the deep neural network is used to predict the transaction data types of credit card.On the one hand,it overcomes the problem of losing a lot of data information in the under-sampling method and the problem of enlarging noise and overfitting in SMOTE and ADASYN when generating new samples.On the other hand,the deep neural network model is applied to the credit card fraud prediction,which expands the application scope of deep learning technology and improves the performance of the model prediction.Finally,based on the real credit card transaction data,this thesis uses the generative adversarial network to generate fraud samples,so that the number of fraud transaction samples and normal transaction samples are approximately balanced,and then uses the deep neural network to build the credit card fraud prediction model.In this study,the model is compared with the common classification algorithm and sampling method in detail,which confirms that the credit card fraud prediction model designed in this study has a good effect,and provides a certain theoretical basis and practical reference for financial institutions to predict credit card fraud.