Network embedding, aiming to learn the low-dimensional representations of nodes in networks, is of paramount importance in many real applications. One basic requirement of network embedding is to preserve the structure and inherent properties of the networks. While previous network embedding methods primarily preserve the microscopic structure, such as the first- and second-order proximities of nodes, the mesoscopic community structure, which is one of the most prominent feature of networks, is largely ignored. In this paper, we propose a novel Modularized Nonnegative Matrix Factorization (M-NMF) model to incorporate the community structure into network embedding. We exploit the consensus relationship between the representations of nodes and community structure, and then jointly optimize NMF based representation learning model and modularity based community detection model in a unified framework, which enables the learned representations of nodes to preserve both of the microscopic and community structures. We also provide efficient updating rules to infer the parameters of our model, together with the correctness and convergence guarantees. Extensive experimental results on a variety of real-world networks show the superior performance of the proposed method over the state-of-the-arts.
Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence
year
2017
pages
203--209
file
Wang et al - Community Preserving Network Embedding.pdf:C\:\\Users\\Admin\\Documents\\Research\\_Paperbase\\Graph Embeddings\\Wang et al - Community Preserving Network Embedding.pdf:application/pdf
%0 Conference Paper
%1 wang_community_2017
%A Wang, Xiao
%A Cui, Peng
%A Wang, Jing
%A Pei, Jian
%A Zhu, Wenwu
%A Yang, Shiqiang
%B Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence
%D 2017
%K Embedding_Algorithm Matrix_Factorization Node_Embeddings
%P 203--209
%T Community Preserving Network Embedding
%X Network embedding, aiming to learn the low-dimensional representations of nodes in networks, is of paramount importance in many real applications. One basic requirement of network embedding is to preserve the structure and inherent properties of the networks. While previous network embedding methods primarily preserve the microscopic structure, such as the first- and second-order proximities of nodes, the mesoscopic community structure, which is one of the most prominent feature of networks, is largely ignored. In this paper, we propose a novel Modularized Nonnegative Matrix Factorization (M-NMF) model to incorporate the community structure into network embedding. We exploit the consensus relationship between the representations of nodes and community structure, and then jointly optimize NMF based representation learning model and modularity based community detection model in a unified framework, which enables the learned representations of nodes to preserve both of the microscopic and community structures. We also provide efficient updating rules to infer the parameters of our model, together with the correctness and convergence guarantees. Extensive experimental results on a variety of real-world networks show the superior performance of the proposed method over the state-of-the-arts.
@inproceedings{wang_community_2017,
abstract = {Network embedding, aiming to learn the low-dimensional representations of nodes in networks, is of paramount importance in many real applications. One basic requirement of network embedding is to preserve the structure and inherent properties of the networks. While previous network embedding methods primarily preserve the microscopic structure, such as the first- and second-order proximities of nodes, the mesoscopic community structure, which is one of the most prominent feature of networks, is largely ignored. In this paper, we propose a novel Modularized Nonnegative Matrix Factorization (M-NMF) model to incorporate the community structure into network embedding. We exploit the consensus relationship between the representations of nodes and community structure, and then jointly optimize NMF based representation learning model and modularity based community detection model in a unified framework, which enables the learned representations of nodes to preserve both of the microscopic and community structures. We also provide efficient updating rules to infer the parameters of our model, together with the correctness and convergence guarantees. Extensive experimental results on a variety of real-world networks show the superior performance of the proposed method over the state-of-the-arts.},
added-at = {2020-02-21T16:09:44.000+0100},
author = {Wang, Xiao and Cui, Peng and Wang, Jing and Pei, Jian and Zhu, Wenwu and Yang, Shiqiang},
biburl = {https://www.bibsonomy.org/bibtex/245f18e044ab2e06dfbbd08d11bf35096/tschumacher},
booktitle = {Proceedings of the {Thirty}-{First} {AAAI} {Conference} on {Artificial} {Intelligence}},
file = {Wang et al - Community Preserving Network Embedding.pdf:C\:\\Users\\Admin\\Documents\\Research\\_Paperbase\\Graph Embeddings\\Wang et al - Community Preserving Network Embedding.pdf:application/pdf},
interhash = {4a5fef8afd2b689561d6b9a3c6bfe079},
intrahash = {45f18e044ab2e06dfbbd08d11bf35096},
keywords = {Embedding_Algorithm Matrix_Factorization Node_Embeddings},
language = {en},
pages = {203--209},
timestamp = {2020-02-21T16:09:44.000+0100},
title = {Community {Preserving} {Network} {Embedding}},
year = 2017
}