Federal learning model optimization method and device

A technology for learning models and optimization methods, applied in the field of deep learning, can solve problems such as the inability to meet the federated learning model and the large amount of user data access, and achieve the effect of protecting security and reducing access.

Pending Publication Date: 2021-07-13
佳讯飞鸿(北京)智能科技研究院有限公司
View PDF1 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this method still has a large amount of access to user data, which cannot meet the needs of effectively and quickly establishing a federated learning model

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
  • Federal learning model optimization method and device
  • Federal learning model optimization method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0030] The technical content of the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0031] Such as figure 1 As shown, the embodiment of the present invention provides a federated learning model optimization method, including the following steps:

[0032] Step S1, establish the agent of each data terminal participating in the federated learning, obtain the local model parameters of each data terminal according to the obtained local training data, and carry out intensive learning training according to the real-time local training data in the storage queue with the preset length obtained, and obtain Local models for individual data terminals.

[0033] In the present invention, the data terminal refers to the self-owned server used by the data provider, and the agent refers to the data processing architecture, and the initial source of the data is the training data.

[0034] Multiple data termi...

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 discloses a federal learning model optimization method and device. The method comprises the following steps: performing model fusion on a local model of each data terminal and a first model obtained by federal learning model parameters in sequence to generate a corresponding initial fusion model; according to a model performance test result of each initial fusion model obtained in the model fusion, determining the weight value of each initial fusion model for each data terminal, and then carrying out comprehensive processing; obtaining the comprehensive weight value of each initial fusion model, and the carrying out sorting; and carrying out model fusion on a set number of initial fusion models which are sorted in the front at a main server side, and obtaining an updated federal learning model, so that, in the federated learning training process, the access of the local model of each data terminal to the user data is reduced, meanwhile, the federated learning model is effectively and quickly established, and then the security of the privacy data of the user is protected.

Description

technical field [0001] The invention relates to a federated learning model optimization method and a corresponding federated learning model optimization device, belonging to the technical field of deep learning. Background technique [0002] As an innovative modeling mechanism, federated learning can provide a machine learning framework to train a unified model for data from multiple parties without compromising the privacy and security of these data. Therefore, applying federated learning to sales, finance, and many other industries can solve the problem that these industries cannot directly aggregate data and train machine learning models caused by factors such as intellectual property rights, privacy protection, and data security. [0003] The invention patent with the patent number ZL 202011044286.5 discloses a method and system for updating model parameters based on federated learning, which can be used for privacy data protection in the process of machine learning. Thi...

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): G06N20/00G06F21/62
CPCG06F21/6245G06N20/00
Inventor 李明春丁晓强单洪政徐震南李如辉
Owner 佳讯飞鸿(北京)智能科技研究院有限公司
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