Cross data segmentation federation learning model method, server and medium

A modeling method and server node technology, which is applied in the field of servers and media, and the field of horizontal data segmentation federated learning modeling methods, can solve the problems of federated learning programs lacking training processes, etc., to protect privacy, improve training efficiency, and increase training speed Effect

Active Publication Date: 2019-01-11
WEBANK (CHINA)
View PDF5 Cites 67 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The main purpose of the present invention is to provide a horizontal data segmentation federated learning modeling method, server and

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
  • Cross data segmentation federation learning model method, server and medium
  • Cross data segmentation federation learning model method, server and medium
  • Cross data segmentation federation learning model method, server and medium

Examples

Experimental program
Comparison scheme
Effect test

no. 1 example

[0123] Based on the first embodiment, the third embodiment of the horizontal data segmentation federated learning modeling method of the present invention is proposed. After step S30, the method of the horizontal data segmentation federated learning modeling method further includes:

[0124] After receiving the end instruction of the local working node and the remote working node to end the federation model training, send the second model parameters to the local working node for the local working node to calculate the second model parameter based on the second model parameter A prediction result, and feed back the first prediction result.

[0125] In this embodiment, online prediction can be performed after the training is completed. After receiving the end instruction of the local working node and the remote working node to end the training, the server node sends the unencrypted second model parameters to the local working node, and the local working node calculates the first...

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

A cross data segmentation federation learning modeling method, a server and a readable storage medium are provided. The method comprises the following steps: the server node sends the first model parameter to each working node, so that each working node obtains the gradient and loss cost by federated model training of its own data to be trained, and feeds back the gradient and loss cost; a gradient and a cost of loss of feedback are received from each working node; the first model parameter is updated based on the gradient and the loss cost to obtain a second model parameter, and whether the second model parameter converges or not is judged. If so, the second model parameter is taken as a standard model parameter. The invention sends model parameters, collects gradient and updates model parameters through server nodes, and the working nodes carry out federated model training at the same time, so that the problem of data leakage does not exist in the model training process according todifferent types of model parameters.

Description

technical field [0001] The invention relates to the field of big data technology, in particular to a horizontal data segmentation federated learning modeling method, a server and a medium. Background technique [0002] The current machine learning solutions for privacy protection mainly remain in theoretical research and academic papers. According to research findings, existing machine learning solutions for privacy protection are limited by technical forms and practical applications, and there are currently no relevant technical applications in the industry. [0003] Moreover, the existing privacy-preserving federated learning schemes often appear in academic papers, but they have insufficient understanding of practical problems, lack of an integrated process from training to prediction, and often only stay at the theoretical research stage. Contents of the invention [0004] The main purpose of the present invention is to provide a horizontal data segmentation federated ...

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
IPC IPC(8): G06F16/2458
Inventor 马国强范涛刘洋陈天健杨强
Owner WEBANK (CHINA)
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap