A network representation learning method and system based on neighbor information
A technology of neighbor information and network representation, applied in the field of network representation learning, can solve problems such as no consideration of neighbor information, and achieve good scalability
Inactive Publication Date: 2018-09-14
ZHEJIANG UNIV
View PDF0 Cites 5 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
The existing work does not take this kind of neighbor information into consideration in the learning of representation vectors.
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0066] The network representation learning system based on neighbor information proposed by the present invention is realized through four modules: data input module, objective function construction module, objective function optimization module, and data output module. The system architecture is as follows: figure 1 shown.
[0067] The flow of the network representation learning method based on the network representation learning system is as follows: figure 1 shown. Since different network structures only have slight differences in the objective function, and the rest are identical, in the following examples only the homogeneous network is taken as an example, and the above method is described in more detail with reference to the accompanying drawings.
[0068] Assuming that an existing user needs to learn a network representation containing 5 nodes, when the user uses the network representation learning system of the present invention to learn network representation, the p...
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
Login to View More Abstract
The invention provides a network representation learning method and system based on neighbor information. The system comprises a data input module, a target function building module, a target functionoptimizing module and a data output module. The data input module inputs a network information data set containing associated information of each network node to the network representation learning system and builds and initializes low-dimension vector representation of each network node. The target function building module builds target functions of the network according to the one-dimension relations of the networks and neighbor information. The target function optimizing module optimizes the target functions by using the stochastic gradient descent algorithm to obtain the optimal vector representation of each network node. The data output module outputs the learnt optimal vector representation of each network node. The invention also provides a method of performing network representation learning by using the network representation learning system. The network representation learning method makes full uses of neighbor information of network nodes, thereby solving the problem of sparsity of one-dimension relations of networks and learning more representative node vectors.
Description
technical field [0001] The invention relates to the technical field of network representation learning, in particular to a network representation learning method and system based on neighbor information. Background technique [0002] Entities in the real world interact with each other to form large-scale complex networks. The traditional network analysis technology regards each network node as a unique symbol. The sparsity problem in this method greatly affects the effect of personalized recommendation and anomaly detection. To overcome the sparsity problem, representation learning methods for complex network analysis are proposed. Network representation learning is to encode the information of nodes in a large-scale network into a low-dimensional space. This low-dimensional representation can be used to judge the distance between nodes, whether there is a relationship, etc., and can also be used as tasks such as classification and clustering. eigenvectors of . [0003] F...
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
Login to View More IPC IPC(8): G06N99/00G06Q50/00
CPCG06Q50/01
Inventor 纪守领杜天宇陈建海
Owner ZHEJIANG UNIV