Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Federation prediction method based on federation learning

A prediction method and federated technology, applied in neural learning methods, prediction, machine learning, etc., can solve problems such as poor prediction results of federated learning, achieve the effects of reducing communication costs, increasing aggregation weights, and improving performance

Active Publication Date: 2021-02-12
GUANGXI NORMAL UNIV
View PDF4 Cites 25 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] What the present invention aims to solve is the problem that the existing federated learning prediction effect is not good, and a federated prediction method based on federated learning is provided

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
  • Federation prediction method based on federation learning
  • Federation prediction method based on federation learning
  • Federation prediction method based on federation learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0032] In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with specific examples.

[0033] For those in the field of federated learning, before federated training, it is necessary to identify federated learning participants and servers, and establish a federated learning environment. Before participating in federated learning, participants need to prepare their own local training data sets to be trained, and at the same time need a local neural network model determined according to the actual application scenario.

[0034] Taking keyboard input prediction as an example, the participants are thousands of mobile devices (mobile phones) participating in federated learning, and the server is Alibaba Cloud or Baidu Cloud. The local training dataset is data of users of input methods such as Gboard who choose to share snippets of text when typing into Google ap...

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 federation prediction method based on federation learning, and the method enables the parameter change of a neural network model updated by a single participant to have onlydirection difference and no size difference through the unitization of a locally updated gradient vector, thereby protecting the data privacy, and avoiding the use of homomorphic and differential privacy or other encryption technologies. Under the condition of not losing data precision, the communication cost between the equipment and the server is greatly reduced. Besides, considering that the difference of data in a federated learning scene is large, the performance of a local participant can be improved by increasing the local information of the data, the uploaded neural network model parameters are clustered by utilizing a k-means algorithm to obtain similar neural network model parameters, the neural network model parameter aggregation weight is improved. The method is more suitable for the data scene of the participant.

Description

technical field [0001] The invention relates to the technical field of federated learning, in particular to a federated prediction method based on federated learning. Background technique [0002] In most industries, due to issues such as industry competition, privacy security, and complex administrative procedures, data is not shared. Even among different departments of the same company, there are many obstacles to achieving centralized data integration. In reality, it is almost impossible to integrate the data scattered in various places and institutions, and the cost required is huge. At the same time, all countries are strengthening the protection of data security and privacy. The new law "General Data Protection Regulation (GDPR)" recently introduced by the European Union shows that the increasingly strict management of user data privacy and security will be a global trend. Aiming at the two major problems of data silos and privacy security, Google proposed a 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
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04G06N3/08G06Q10/04G06N20/00
CPCG06N3/04G06N3/08G06Q10/04G06N20/00G06F18/23213
Inventor 李先贤段锦欢王金艳
Owner GUANGXI NORMAL UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com