An Account Identity Verification Model Based On The Pseudo-siamese Network

With the rapid development of Internet,Internet payment becomes more and more convenient.Today,when people are highly dependent on online payment,there are also huge risks of online account funds.Such as account theft,fictitious transactions,identity information stolen and malicious network lending,the use of false number after registration of bulk orders,consumer finance platform bad cash,using Trojan virus to steal mobile phone bank money and so on...In the presence of such huge risks,the anti-fraud ability of digital fintech has been pushed to the forefront in the current era of rapid development offintech.The verification of user's identity is the top priority in the risk control gateway.The existing verification of user's identity is mainly based on account password verification,or verification code identification and so on.This paper innovatively proposes a new way to verify the user's identity based on the user's own biometrics.Compared with the other methods,based on the biological characteristics of non-inductive check method is a comparatively perfect authentication way,not only can achieve higher accuracy,at the same time,has the very high intensity risk control confrontational,it is difficult to be cracked and the solutions in the process of the application to run at the same time,in the absence of user attention to authenticate user's identity,also made it greatly retained the user experience.In addition to the innovation in engineering technology,this paper also has innovation in algorithm.Based on the verification characteristics of biological characteristics,this paper proposes a verification model,which is based on the pseudo-Siamese network It will also be compared with the latest approaches to sequential data in the areas of speech,natural language processing and biological signal analysis.The behavior recognition method in this paper is fully tested on a real data set named BRSB.It consists of more than 2000 interactive conversations with 131 users and more than 26,000 touch interactions.The experimental results prove the effectiveness and superiority of the method.