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

Graph neural network training method, client device and system

A neural network training and client device technology, applied in the field of graph neural networks, can solve problems such as low work efficiency, reduce workload and alleviate low work efficiency.

Active Publication Date: 2022-08-05
NETEASE (HANGZHOU) NETWORK CO LTD
View PDF12 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a graph neural network training method, client device and system, to alleviate the technical problem of low work efficiency in the graph neural network training method in the prior art

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
  • Graph neural network training method, client device and system
  • Graph neural network training method, client device and system
  • Graph neural network training method, client device and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0060] A graph neural network training method provided by an embodiment of the present invention is applied to a client device for graph neural network training, and the client device is communicatively connected to a server device for graph neural network training, such as figure 1 As shown, the method includes the following steps:

[0061] Step S11, in response to the user's environment creation operation, a graph training environment corresponding to the environment creation operation is created using the resources of the server device.

[0062] In the embodiment of the present invention, to train the graph neural network to be trained, the user only needs to interact with the client device trained by the graph neural network. End devices are used to provide the resources required for graph neural network training.

[0063] First, the user sends a request for a graph neural network training environment creation operation to the client device, and then the client device res...

Embodiment 2

[0107] An embodiment of the present invention also provides a graph neural network training client device, the client device is connected to the graph neural network training server device in communication, and is mainly used to perform the graph neural network training provided in the first embodiment above method, the following is a specific introduction to the client device for training the graph neural network provided by the embodiment of the present invention.

[0108] Figure 5 It is a functional module diagram of a client device for graph neural network training provided by an embodiment of the present invention. The client device includes: an environment creation module 10, a model creation module 20, a setting module 30, a training module 40, and a determination module 50, in:

[0109] The environment creation module 10 is configured to respond to the user's environment creation operation, and use the resources of the server device to create a graph training environ...

Embodiment 3

[0137] The embodiment of the present invention also provides a graph neural network training system, such as Image 6 As shown, the graph neural network training system includes the client device 1 trained by the graph neural network in the first embodiment, and further includes: the server device 2 trained by the graph neural network;

[0138] Wherein, the client device 1 is connected in communication with the server device 2;

[0139] The server device 2 is used to provide resources required for training the graph neural network.

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 present invention provides a graph neural network training method, client device and system, which relate to the technical field of graph neural networks. The method is applied to a graph neural network training client device. First, a graph is created in response to a user's environment creation operation. The training environment of the neural network, and then responds to the user's model creation operation to create the neural network to be trained, and then responds to the user's setting operation to set the hyperparameters of the neural network to be trained, and then responds to the user's training operation. The graph neural network to be trained corresponding to the hyperparameter combination is trained, and the optimal hyperparameter combination corresponding to the graph data training set is finally determined. On the basis of making full use of the graph data structure, the present invention quickly and automatically obtains the graph to be trained. The optimal hyperparameter combination of the neural network for the graph data training set reduces the workload of the user for parameter adjustment, thereby alleviating the technical problem of low work efficiency of the graph neural network training method in the prior art.

Description

technical field [0001] The present invention relates to the technical field of graph neural networks, and in particular, to a graph neural network training method, client device and system. Background technique [0002] Graph Neural Network (GNN) is a method of processing graph data, mainly a method of processing graph domain information based on deep learning, which can be processed according to the characteristics of graph data, because graph data is irregular (non-European) Yes, the graph neural network needs to regularize the graph data before learning or during the learning process, and process it into training samples of the same dimension on the basis of retaining the graph structure information as much as possible. The commonly used method is the control of neighboring nodes, but Neighbor node control not only solves the problem of data regularization, but also introduces new hyperparameters, such as the number of neighbor nodes and sorting rules, which makes the wor...

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): G06N3/04G06N3/08
CPCG06N3/08G06N3/045
Inventor 朱钰森尚书胡志鹏
Owner NETEASE (HANGZHOU) NETWORK CO LTD