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

Deep attention embedded graph clustering method with smooth structure

An attention and graph clustering technology, applied in the field of graph data processing, can solve problems such as clustering performance degradation, achieve the effect of eliminating instability and solving performance degradation problems

Pending Publication Date: 2022-08-02
湖北楚天高速数字科技有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, due to the introduction of target distribution and soft distribution distribution in the self-optimization process, the fixed-period distribution update strategy makes the clustering performance easy to degrade
In this regard, it is still a challenge to find a model that is highly adaptable, stable and suitable for graph clustering tasks

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
  • Deep attention embedded graph clustering method with smooth structure

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0033] like figure 1 The implementation process of the deep attention embedded graph clustering method with smooth structure provided by the embodiment of the present invention is shown. For convenience of description, only the part related to the embodiment of the present invention is shown. Details are as follows:

[0034] Step 1. Preprocess the graph data, construct the basic attribute graph G=(V, E, X, A), and set the structure of the basic graph attention coding network, including the number of network layers, the dimension of the hidden layer, Output dimension and initial clustering accuracy Acc;

[0035] Step 2. Input the graph structure and node attribute information to define the loss term L of the graph coding network r with L s , some hyperparameters are pre-set by a multi-group cross-validation method, including the penalty term coefficient: ρ l with λ l ;

[0036] Step 3. Integrate the structural reconstruction loss L r and the smoothness constraint L s , ...

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 is applicable to the field of graph data processing, provides a deep attention embedded graph clustering method with a smooth structure, and provides a novel graph clustering method which can effectively eliminate instability of a reconstructed graph structure by simultaneously applying an attention mechanism and a smooth constraint on a graph self-encoding network. The invention further provides a self-optimization clustering module, and the problem of performance degradation caused by target distribution updating is effectively solved.

Description

technical field [0001] The invention belongs to the field of graph data processing, and in particular relates to a deep attention embedded graph clustering method with smooth structure. Background technique [0002] In the past few decades, continuous breakthroughs in deep learning techniques have greatly changed the learning paradigm in various fields, and have also achieved state-of-the-art performance in many important tasks, including classification and clustering tasks. Among them, the deep graph clustering method represented by graph representation learning has attracted great attention, and its goal is to capture the relationship between nodes by jointly modeling node attributes and graph structure, and divide the graph into several external Distinctive, interconnected communities or groups. However, given the high sparsity of the graph structure, many challenges remain in the clustering task of unsupervised graphs. In recent years, graph neural network GNN has achi...

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): G06V10/762G06V10/774G06V10/82G06N3/04G06K9/62
CPCG06V10/763G06V10/774G06V10/82G06N3/045G06F18/23213
Inventor 阮一恒朱晨露高源张立杰
Owner 湖北楚天高速数字科技有限公司
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