Data Feature Extraction Method, Device, and Computer-Readable Storage Medium

Abstract

The embodiment of the invention discloses a data feature extraction method and device and a computer readable storage medium which are applied to the computer field. The data feature extraction methodcomprises the steps of extracting features of behavior data in each sample through a preset first neural network, and inputting the extracted features into a preset second neural network; performingadversarial training on the preset first neural network and the preset second neural network; stopping the training when the trained second neural network outputs a result value belonging to a presetvalue range; and extracting features of target behavior data through the trained first neural network, wherein the target behavior data belongs to the same data category as the behavior data, therebynot only being capable of ensuring the features not to be interfered by the changes in data distribution, but also simplifying the process of extracting features with sample distribution invariance compared with the existing sorting method, shortening the operation time, and reducing the lone-term occupation for computing resources.

Description

technical field [0001] Embodiments of the present invention belong to the field of computer technology, and in particular relate to a data feature extraction method, device, and computer-readable storage medium. Background technique [0002] Online payment is a form of electronic payment, which is an instant payment method through the payment interface with the bank provided by a third party. Due to the convenience of online payment and the quickness of funds arriving in the account, more and more banks, merchants, and users choose online payment. With the development of online payment, online payment platforms can also have functions such as applying for loans, credit cards, or purchasing insurance and wealth management products. Before opening these functions, banks or payment platforms need to conduct credit scores on users. Users with low credit are prone to late repayment or non-repayment, so credit review is particularly important. [0003] In the prior art, features...

AI Technical Summary

Problems solved by technology

[0005] Embodiments of the present invention provide a data feature extraction method, device, and computer-readable storage medium, aiming to solve the huge amount of calculation and time-consuming process caused by the process of extracting features through the sorting method, which in turn leads to long-term occupation of a large number of calculations resource problem

Images