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

Method and device for training graph neural network based on privacy protection

A neural network, privacy protection technology, applied in the computer field, can solve problems such as graph data information leakage

Active Publication Date: 2021-08-20
ALIPAY (HANGZHOU) INFORMATION TECH CO LTD
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, if the model is stolen, the large-scale graph data information used to train the model is also at risk of being leaked

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 and device for training graph neural network based on privacy protection
  • Method and device for training graph neural network based on privacy protection
  • Method and device for training graph neural network based on privacy protection

Examples

Experimental program
Comparison scheme
Effect test

Embodiment approach

[0114] According to one embodiment, the sampling unit 53 is configured to: normalize the matching degrees corresponding to the respective neighboring nodes to obtain corresponding matching probabilities; and sample the respective neighboring nodes according to the matching probabilities .

[0115] According to another embodiment, the sampling unit 53 includes (not shown):

[0116] The first probability determination module is configured to determine a first sampling probability of the second node being sampled based on the index mechanism of differential privacy, according to the first privacy budget and the matching degree between the second node and the first node ;

[0117] The neighbor sampling module is configured to sample each neighbor node according to the first sampling probabilities respectively corresponding to each neighbor node in the first neighbor node set.

[0118] Further, in one embodiment, the neighbor sampling module is configured to: perform a predetermi...

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 embodiment of this specification provides a method and device for training a graph neural network based on privacy protection. The method includes first obtaining an original relational network graph, and any first node in the graph has a corresponding neighbor node set. For any second node in the neighbor node set, the node information of the second node, the node information of the first node, and the connection information between the second node and the first node are input into the multilayer neural network, and the second node and the first node are obtained. Matching degree of a node. Then, according to the matching degree corresponding to each neighbor node in the neighbor node set, the neighbor node set is sampled to obtain the sampled neighbor node set of the first node. Then, based on the sampling neighbor node sets corresponding to each node in the original graph, a sparse relational network graph is formed. Then, based on the sparse relational network graph, the graph neural network is trained.

Description

technical field [0001] One or more embodiments of this specification relate to the field of computer technology, and in particular to a method and device for training a graph neural network based on privacy protection. Background technique [0002] Relational network graphs have recently become a central area of ​​machine learning. Data mining and machine learning based on relational network graphs have played more and more value in many fields. For example, the structure of social networks can be understood by predicting potential connections, fraud detection can be performed based on graph structures, consumer behavior of e-commerce users can be understood or real-time recommendations can be made, etc. [0003] At the same time, people's emphasis on privacy is also increasing day by day. There is a large amount of information contained in the relationship network graph. If various artificial intelligence and machine learning (AI / ML) models using graph information are not...

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 Patents(China)
IPC IPC(8): G06F21/62G06N3/08G06N3/04
CPCG06F21/6245G06N3/084G06N3/045G06F18/2415
Inventor 熊涛
Owner ALIPAY (HANGZHOU) INFORMATION TECH CO LTD