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

Missing feature re-representation method and system based on graph convolutional network

A convolutional network, network technology

Pending Publication Date: 2021-06-25
INST OF COMPUTING TECH CHINESE ACAD OF SCI
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to overcome the difficulty of model training due to the lack of features in the process of machine learning. On this issue, a method for re-representing data missing features based on graph convolutional networks is proposed.

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
  • Missing feature re-representation method and system based on graph convolutional network
  • Missing feature re-representation method and system based on graph convolutional network
  • Missing feature re-representation method and system based on graph convolutional network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0035] When the inventor was conducting research on missing features, he found that the processing methods for missing data in the prior art ignored the correlation between samples of missing data and other samples. After research, the inventor found that the collected samples of original data The correlation between them makes the problem of missing data features can be solved by making full use of the information of other samples to re-express the missing data.

[0036] The present invention proposes a method for re-representing missing features based on a graph convolutional network. The method proposed by the present invention includes two stages: the construction of the graph network and the training of the graph network. In the construction phase of the graph network: firstly, a graph network is constructed by extracting features from the original data set of the machine learning task, and each sample in the original data set is used as a node in the graph network, and t...

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 provides a missing feature re-representation method and system based on a graph convolutional network. The method comprises the steps: extracting the features of a training sample with a marked category, taking the features of the training sample as nodes, obtaining the similarity relation between the nodes through distance measurement, constructing a connection edge between the nodes according to the similarity relation, and carrying out the re-representation of the missing features of the training sample. Obtaining a graph network of the training sample; Training a graph convolutional network according to the feature information of the adjacent samples in the graph network and the annotation categories corresponding to the nodes in the graph to obtain a feature re-representation model, inputting the features of the to-be-classified samples into the feature re-representation model, reconstructing the features of the to-be-classified samples, and classifying the reconstructed features to obtain a classification result. And obtaining a classification result of the to-be-classified sample.

Description

technical field [0001] The invention relates to the field of machine learning, in particular to a method for re-expressing missing features based on graph convolutional networks. As a method for dealing with abnormal data such as missing features in machine learning, this method can effectively solve the problem of model failure caused by missing features. Background technique [0002] The current machine learning model method needs to keep the feature dimension consistent, but in the real environment, there is a problem of missing source data, such as unstable wireless signals, the characteristics of the sensor itself, and severe changes in the highly dynamic environment will cause data loss situation, and the severity increases with the continuous expansion of the range of environments to which the model is adapted. The lack of data will lead to problems such as increased difficulty in extracting features, thereby reducing the performance of machine learning models. [0...

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): G06N3/04G06N3/08
CPCG06N3/08G06N3/045
Inventor 蒋鑫龙陈益强沈鸿张忠平王永斌刘廉如
Owner INST OF COMPUTING TECH CHINESE ACAD OF SCI
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