Anti-fraud model training method, system and anti-fraud method based on transfer learning

Abstract

The present application provides an anti-fraud model training method, a system and an anti-fraud method based on migration learning in electronic banking, wherein, the anti-fraud model training methodof electronic banking comprises the following steps: acquiring first characteristic data of an auxiliary business scenario and second characteristic data of a target business scenario; Determining afirst value of a parameter in the anti-fraud model based on the first characteristic data and the second characteristic data; Determining a feature probability value based on the first value, the anti-fraud model, and the second feature data; Determining a second value of a parameter in the anti-fraud model based on the second characteristic data and the characteristic probability value, and updating the anti-fraud model using the second value. The present application introduces a migration learning strategy method into an electronic banking anti-fraud system, and simultaneously takes into account the operation characteristic data of a target business scenario and an auxiliary business scenario, so that the fraud identification accuracy of the anti-fraud model is greatly improved.

Description

technical field [0001] The present application relates to the field of computer information technology, in particular to a transfer learning-based anti-fraud model training method, system and anti-fraud method. Background technique [0002] The rapid development of the Internet and the popularity of smart terminals have brought great convenience to people when using e-banking to handle balance inquiries, transfers, shopping payments, and financial management. to the risk of malicious infringement. [0003] Surveys show that cybercrime is increasingly sophisticated and penetrates into different industries, bringing economic losses of up to 445 billion U.S. dollars to the world every year. Moreover, fraud has developed into a black industry chain with strict organization and clear professional division of labor, which has brought severe challenges to banks' development of online financial business. [0004] At present, the prevention and control strategies of electronic bank...

AI Technical Summary

Problems solved by technology

And fraud has developed into a black industry chain with strict organization and clear professional division of labor, which has brought severe challenges to the development of online financial services for banks

[0004] At present, the prevention and control strategies of electronic banking anti-fraud systems in the industry generally use two implementation schemes: one is based on expert rules, and pure expert rules are difficult to meet the needs of current electronic banking anti-fraud systems; Machine learning of the risk characteristics of the whole process and offline analysis of historical data

[0005] The existing machine learning-based method is to directly apply the machine learning anti-fraud model to the current scenario without modification. Although it avoids human subjectivity to a certain extent compared with expert rules, it ignores different business scenarios. Due to the complexity and diversity of the operation process and operation characteristic data of the e-banking business system, direct anti-fraud model training without distinguishing business scenarios will bring difficulties to machine learning

Images