TY - JOUR
T1 - Spectral embedding network for attributed graph clustering
AU - Zhang, Xiaotong
AU - Liu, Han
AU - Wu, Xiao Ming
AU - Zhang, Xianchao
AU - Liu, Xinyue
N1 - Funding Information:
This work is supported by the Fundamental Research Funds for the Central Universities, China (No. DUT20RC(3)066 , No. DUT20RC(3)040 ), and National Natural Science Foundation of China (No. 61632019 , No. 61876028 , No. 61972065 ).
Publisher Copyright:
© 2021 Elsevier Ltd
PY - 2021/10
Y1 - 2021/10
N2 - Attributed graph clustering aims to discover node groups by utilizing both graph structure and node features. Recent studies mostly adopt graph neural networks to learn node embeddings, then apply traditional clustering methods to obtain clusters. However, they usually suffer from the following issues: (1) they adopt original graph structure which is unfavorable for clustering due to its noise and sparsity problems; (2) they mainly utilize non-clustering driven losses that cannot well capture the global cluster structure, thus the learned embeddings are not sufficient for the downstream clustering task. In this paper, we propose a spectral embedding network for attributed graph clustering (SENet), which improves graph structure by leveraging the information of shared neighbors, and learns node embeddings with the help of a spectral clustering loss. By combining the original graph structure and shared neighbor based similarity, both the first-order and second-order proximities are encoded into the improved graph structure, thus alleviating the noise and sparsity issues. To make the spectral loss well adapt to attributed graphs, we integrate both structure and feature information into kernel matrix via a higher-order graph convolution. Experiments on benchmark attributed graphs show that SENet achieves superior performance over state-of-the-art methods.
AB - Attributed graph clustering aims to discover node groups by utilizing both graph structure and node features. Recent studies mostly adopt graph neural networks to learn node embeddings, then apply traditional clustering methods to obtain clusters. However, they usually suffer from the following issues: (1) they adopt original graph structure which is unfavorable for clustering due to its noise and sparsity problems; (2) they mainly utilize non-clustering driven losses that cannot well capture the global cluster structure, thus the learned embeddings are not sufficient for the downstream clustering task. In this paper, we propose a spectral embedding network for attributed graph clustering (SENet), which improves graph structure by leveraging the information of shared neighbors, and learns node embeddings with the help of a spectral clustering loss. By combining the original graph structure and shared neighbor based similarity, both the first-order and second-order proximities are encoded into the improved graph structure, thus alleviating the noise and sparsity issues. To make the spectral loss well adapt to attributed graphs, we integrate both structure and feature information into kernel matrix via a higher-order graph convolution. Experiments on benchmark attributed graphs show that SENet achieves superior performance over state-of-the-art methods.
KW - Attributed graph clustering
KW - Graph structure improvement
KW - Kernel matrix learning
KW - Spectral embedding network
UR - http://www.scopus.com/inward/record.url?scp=85107769349&partnerID=8YFLogxK
U2 - 10.1016/j.neunet.2021.05.026
DO - 10.1016/j.neunet.2021.05.026
M3 - Journal article
AN - SCOPUS:85107769349
SN - 0893-6080
VL - 142
SP - 388
EP - 396
JO - Neural Networks
JF - Neural Networks
ER -