Credit risk prediction model training method and credit risk prediction method

Abstract

The invention provides a credit risk prediction model training method and a credit risk prediction method, and the method comprises the steps: obtaining credit sample data, and repeatedly executing the following steps until a convergent credit risk prediction model is obtained: inputting sample source data into a convolutional neural network model to obtain a first feature result output by the convolutional neural network model, the first feature result is a continuous evaluation value representing the credit risk occurrence probability corresponding to the sample source data; the first feature result is input into a binary random model, a second feature result output by the binary random model is obtained, and the second feature result is used for randomly generating a dichotomy value representing whether the credit risk corresponding to the sample source data occurs or not; based on a preset loss function, generating a loss function value corresponding to the sample source data through the second feature result and the corresponding label value, and performing parameter adjustment on the convolutional neural network model through the loss function value; the problem of gradient disappearance is avoided, and the training efficiency of the model is improved.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, in particular to a training method of a credit risk prediction model and a credit risk prediction method. Background technique [0002] Credit business is an important business in the Internet finance industry. With the continuous development of the domestic Internet finance industry and the advent of the era of big data, the number of loans and the amount of loans have increased significantly. How to effectively assess the credit risk of borrowers has become a financial It is one of the key links for the sustainable and healthy development of the industry. The current technical solution to realize credit risk prediction is to treat it as a binary classification problem, and evaluate whether there will be credit risk based on the credit information of loan users and other user characteristics, that is, repayment on time or overdue repayment, so as to realize the loan bef...

AI Technical Summary

Problems solved by technology

[0005] This application provides a training method of a credit risk prediction model and a credit risk prediction method to solve problems affecting the accuracy and effectiveness of credit risk prediction

Images