Unlock instant, AI-driven research and patent intelligence for your innovation.

Credit business recommendation method based on federal learning

A business recommendation and credit technology, applied in the information field, can solve problems such as the inability of enterprises to recommend the best credit business products, and achieve the effect of recommending credit business products accurately, improving accuracy, and improving privacy and security.

Active Publication Date: 2022-07-22
浙江数秦科技有限公司
View PDF12 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, its technical solutions cannot recommend the best credit business products for enterprises

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Credit business recommendation method based on federal learning
  • Credit business recommendation method based on federal learning
  • Credit business recommendation method based on federal learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0047] A federated learning-based credit business recommendation method, please refer to the appendix figure 1 ,include:

[0048]Step A01) Establish a trusted server, and the trusted server establishes a credit business display page to display credit business products of several financial institutions for the enterprise to choose. The types of credit business products include credit loans, mortgage loans, letters of guarantee and discounted bills. There are long-term loans, medium and long-term loans and short-term loans in time. In addition, the interest rate calculation methods and repayment methods of credit products are different from each other. In this regard, enterprises need to compare among many credit products and choose the most suitable credit product for their own situation. Enterprises need to spend a lot of time to obtain and compare credit products, which is very time-consuming and labor-intensive. A trusted server is established to centrally display credit ...

Embodiment 2

[0082] A method for recommending a credit business based on federated learning, this embodiment provides specific protection for enterprise information on the basis of the first embodiment. In this embodiment, when the enterprise requests the trusted server to view the credit service display page, the trusted server generates a temporary asymmetric encryption key pair for the enterprise, which is recorded as a public key and a private key respectively. The public key is sent to the enterprise, and the enterprise encrypts the enterprise information with the public key and sends it to the trusted server. The trusted server decrypts the enterprise information with the private key. Then, the enterprise information and the credit business classification information of each credit business product are input into the final neural network model after preprocessing to obtain the recommendation level.

[0083] Please see attached Figure 5 , the process of the trusted server generatin...

Embodiment 3

[0108] A method for recommending a credit business based on federated learning. On the basis of the first embodiment, this embodiment further provides a specific method for exchanging obfuscation values ​​with higher security. Please see attached Figure 8 , several financial institutions generate mutually offset obfuscated values ​​including:

[0109] Step F01) The financial institution generates a random positive odd number, calculates the cosine value of the random positive odd number, and discloses the cosine value;

[0110] Step F02) Each financial institution is paired with other financial institutions in turn, and the cosine value of the product of random positive and odd numbers generated by the paired two financial institutions is calculated using the double angle formula of the cosine function;

[0111] Step F03) Take the first preset decimal of the cosine value as the absolute value of the confusion value, and determine the positive and negative signs of the confus...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to the technical field of information, in particular to a federated learning-based credit business recommendation method, which comprises the following steps of: establishing a trusted server, and displaying a credit business product; the financial institution collects credit business data; manually marking recommendation levels to form sample data, and respectively establishing and training neural network models; the model parameters are encrypted and then sent to a trusted server; the trusted server fuses the model parameters; the financial institution updates the neural network model, uses sample data for training, encrypts a loss value and then sends the loss value to the trusted server; the trusted server calculates a global loss value, and if the global loss value reaches preset accuracy, the model parameters are fused again to complete federated learning; and the enterprise provides enterprise information for the trusted server, inputs the final neural network model to obtain output of the final neural network model, and sorts the output in a descending order as a credit business recommendation result of the enterprise. The method has the substantive effect that credit business product recommendation is realized on the premise of ensuring data privacy.

Description

technical field [0001] The invention relates to the field of information technology, in particular to a credit business recommendation method based on federated learning. Background technique [0002] Loans refer to the general term for loans, discounts, overdrafts and other lending funds. Banks release the centralized currency and monetary funds through loans, which can meet the needs of supplementary funds for social expansion of reproduction and promote economic development. At the same time, banks can also obtain loan interest income and increase their own accumulation. According to the loan term, it is divided into: medium and long-term loans, loan terms of more than 5 years, medium-term loans, loan terms of more than 1 year, less than 5 years, short-term loans, loan terms of less than 1 year, overdraft, loans with no fixed term. According to the loan guarantee method, it is divided into: credit loan, guaranteed loan, guaranteed loan, mortgage loan, pledge loan and bil...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06Q40/02H04L9/40G06N3/08
CPCG06N3/08H04L63/062H04L63/0442H04L2209/56G06Q40/03
Inventor 俞学劢张金琳
Owner 浙江数秦科技有限公司