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

Method for protecting privacy in federated learning prediction stage based on PSI technology

A federal, phased technology, applied in the field of information security, can solve problems such as no research

Active Publication Date: 2020-06-09
百融云创科技股份有限公司
View PDF3 Cites 18 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although PSI technology is relatively mature, there is no relevant research on how to use it in the prediction stage of federated learning

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
  • Method for protecting privacy in federated learning prediction stage based on PSI technology
  • Method for protecting privacy in federated learning prediction stage based on PSI technology

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] Let the machine learning algorithm used in the present invention be logistic regression, the data feature of Alice is Bob's data features are where d A and d B are the number of features owned by Alice and Bob respectively. The partial models obtained by Alice and Bob through federated learning are respectively with where θ -1 is the intercept. The prediction function of logistic regression is in For specific sample data, the first half Owned by Alice, second half owned by Bob. Then Alice performs the prediction according to the following steps:

[0033] 1. For all id∈ID B , Bob calculates Form a new data set and record it as Predict B ={(id,score B,id )|id∈ID B};

[0034] 2. Alice and Bob execute the improved OT protocol-based PSI to obtain new data set eID respectively A ={eid A,id |id∈ID A} and eP B ={(eid B,id ,score B,id )|id∈ID B};

[0035] 3. For each (eid B,id ,score B，id )∈eP B , Bob uses the key derivation function KDF to g...

Embodiment 2

[0043] Same as the symbol used in Embodiment 1, the prediction is also made using logistic regression, the only difference is that the PSI based on the public key is used in this embodiment. For this reason, some other symbols need to be agreed upon. Assume that this implementation case is based on elliptic curves, and q is the order of elliptic curves. Represents the set {1,2,...,q-1}, H 1 and H 2 is a hash function, where H 1 Messages can be mapped to points on an elliptic curve. Specific steps are as follows:

[0044] 1. Precomputation, for all id∈ID B , Bob randomly chooses calculate s id =H 2 (a*H 1 (id)), and all s id ,id∈ID B send to Alice;

[0045] 2. For all id ∈ ID A , Alice randomly chooses calculate t id =b*H 1 (id), and t id ,id∈ID A send to Bob;

[0046] 3. Bob performs the following steps:

[0047] 3.1. For all id∈ID B , according to Alice's needs, Bob calculates

[0048] 3.2. For all id∈ID B , using the key derivation function to comp...

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 method for protecting privacy in a federated learning prediction stage based on a PSI technology. The method comprises the following steps that firstly, a prediction serviceparty calculates a prediction result of a model of the prediction service party, then the prediction service party and the prediction service party execute an improved PSI protocol, a part of model calculation results of the prediction service party are encrypted, a prediction demand party decrypts the calculation result of a data provider in combination with data of the prediction demand party, and finally a prediction result of a keyword id shared by the prediction service party and the prediction demand party is obtained. According to the method, the PSI technology is utilized, the data areencrypted through the key derivation function, the privacy protection requirement in the federated learning prediction stage is met, the last link of federated learning privacy security protection isbroken through, and landing of more application scenes of federated learning is promoted.

Description

technical field [0001] The invention relates to the fields of information security and artificial intelligence, in particular to the application of PSI technology in federated learning to realize privacy protection in the prediction stage of federated learning, and is a method for protecting privacy in the predicted stage of federated learning based on PSI technology. Background technique [0002] The rapid development of emerging technologies such as big data, cloud computing, and the Internet of Things has led to an explosive growth of data, which is in the hands of different entities. On the one hand, with the promulgation of laws and regulations such as my country's "Network Security Law" and the EU's "General Data Protection Regulation" (GDPR), governments of various countries will have stricter requirements for user data privacy protection. Bottlenecks have emerged in the use and analysis of data; on the other hand, various entities that hold a large amount of data hope...

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): G06F21/62G06F21/60G06N20/00
CPCG06F21/6245G06F21/602G06N20/00
Inventor 张韶峰单进勇
Owner 百融云创科技股份有限公司
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