Third-party platform payment fraud behavior online detection method based on LSTM

Abstract

The invention provides a third-party platform payment fraud behavior online detection method based on LSTM. The method comprises the following steps: acquiring data and preprocessing the acquired data; dividing the data into a training set and a test set; extracting main features according to the cumulative variance contribution rate CPV of the feature vectors; constructing an LSTM unit and an LSTM model, and using a mean square error as a loss function to update model parameters; and updating the weight and bias in the LSTM model by adopting an Adam gradient descent algorithm, optimizing themodel by using the test set, and finally deploying the optimized LSTM model to the rear end of the third-party payment platform to perform online anomaly detection monitoring on the account behavior of the customer. By implementing the technical scheme of the invention, the early warning result accuracy is high, the fraud identification capability is strong, the financial fraud risk is reduced, the timely system early warning can be sent out, and the account operation transaction can be blocked according to the risk level.

Description

technical field [0001] The invention relates to the technical field of risk control in the Internet financial third-party payment industry, in particular to an LSTM-based online detection method for third-party platform payment fraud. Background technique [0002] Under the concept of "Internet +", third-party mobile payment represented by Alipay and WeChat payment has developed rapidly. According to relevant reports, the transaction scale of third-party mobile payment reached 226.2 trillion yuan in 2019, a year-on-year increase of 18.7%. While mobile payment has brought great convenience to life, it also brings huge risks, such as a large number of fraudulent activities such as counterfeit applications, stolen funds, and account theft. The existing anti-fraud technologies of third-party payment platform companies are mostly based on big data credit evaluation systems and pre-defined anti-fraud rules. There is still a certain lag in the practice, and it is impossible to de...

AI Technical Summary

Problems solved by technology

While mobile payment brings great convenience to life, it also brings huge risks. There are a large number of fraudulent activities such as counterfeit applications, fund theft, and account theft.

The existing anti-fraud technologies of third-party payment platform companies are mostly applied to credit evaluation systems based on big data and pre-defined anti-fraud rules. There is still a certain lag in the practice, and it is impossible to detect and identify new types of fraud in real time and cause financial losses

Images