Trusted electronic transaction release mechanism based on generative adversarial network

Abstract

The invention discloses a trusted electronic transaction release mechanism based on a generative adversarial network. The method is characterized by comprising, the transaction distribution of normalusers is learnt by using the generative adversarial network by using a large amount of non-fraudulent data based on a reverse idea; the obtained discrimination model can better identify non-fraud data, so that a release mechanism can be adopted for credible electronic transactions, and the remaining data which cannot be discriminated are delivered to the prediction model for prediction, so that alarge amount of time can be saved.

Description

technical field [0001] The invention relates to anti-fraud detection of Internet financial network payment. Background technique [0002] The mobile Internet is a double-edged sword. While it brings convenience to people's lives, it also brings various hidden dangers. For example, the payment platform for online transactions allows people to shop without leaving home or even anytime, anywhere. And payment, but this convenience and speed also give some unscrupulous attackers an opportunity. Attackers steal users' account information, steal users' personal privacy information, and even pretend to be users themselves to conduct transactions or transfers to complete fraud. Therefore, in order to effectively protect the personal interests of users and companies, it is necessary to establish an effective network payment fraud detection system. [0003] In the transaction data of Internet finance, fraudulent data usually account for a very small proportion, because most of the use...

AI Technical Summary

Problems solved by technology

[0003] In the transaction data of Internet finance, fraudulent data usually account for a very small proportion, because most of the user's transactions are normal, and only a very small part of them is fraudulent, and most of the methods in this field It is to catch thieves, that is, to predict fraudulent data, but the prediction process is usually very time-consuming. We know that time is loss. The faster you can find fraudulent transactions, the sooner you can avoid user losses. The prediction model is more time-consuming. It is almost impossible to train a model that can predict fraudulent transactions when the fraudulent and non-fraudulent data are extremely imbalanced.

Images