HCL: Improving Graph Representation with Hierarchical Contrastive Learning

被引:0
作者
Wang, Jun [1 ]
Li, Weixun [1 ]
Hou, Changyu [1 ]
Tang, Xin [2 ]
Qiao, Yixuan [1 ]
Fang, Rui [2 ]
Li, Pengyong [3 ]
Gao, Peng [1 ]
Xie, Guotong [1 ,4 ,5 ]
机构
[1] Ping Healthcare Technol, Beijing, Peoples R China
[2] Ping Property & Casualty Insurance Co, Shenzhen, Peoples R China
[3] Xidian Univ, Sch Comp Sci & Technol, Xian, Peoples R China
[4] Ping Hlth Cloud Co Ltd, Shenzhen, Peoples R China
[5] Ping Int Smart City Technol Co Ltd, Shenzhen, Peoples R China
来源
SEMANTIC WEB - ISWC 2022 | 2022年 / 13489卷
关键词
Data mining; Graph learning; Contrastive learning;
D O I
10.1007/978-3-031-19433-7_7
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Contrastive learning has emerged as a powerful tool for graph representation learning. However, most contrastive learning methods learn features of graphs with fixed coarse-grained scale, which might underestimate either local or global information. To capture more hierarchical and richer representation, we propose a novel Hierarchical Contrastive Learning (HCL) framework that explicitly learns graph representation in a hierarchical manner. Specifically, HCL includes two key components: a novel adaptive Learning to Pool (L2Pool) method to construct more reasonable multi-scale graph topology for more comprehensive contrastive objective, a novel multi-channel pseudo-siamese network to further enable more expressive learning of mutual information within each scale. Comprehensive experimental results show HCL achieves competitive performance on 12 datasets involving node classification, node clustering and graph classification. In addition, the visualization of learned representation reveals that HCL successfully captures meaningful characteristics of graphs.
引用
收藏
页码:108 / 124
页数:17
相关论文
共 48 条
  • [21] Lee J, 2019, PR MACH LEARN RES, V97
  • [22] Li PY, 2021, PROCEEDINGS OF THE THIRTIETH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, IJCAI 2021, P2694
  • [23] Liu X., 2020, arXiv
  • [24] Geometric deep learning on graphs and manifolds using mixture model CNNs
    Monti, Federico
    Boscaini, Davide
    Masci, Jonathan
    Rodola, Emanuele
    Svoboda, Jan
    Bronstein, Michael M.
    [J]. 30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017), 2017, : 5425 - 5434
  • [25] Kipf TN, 2016, Arxiv, DOI arXiv:1611.07308
  • [26] Narayanan A, 2017, Arxiv, DOI arXiv:1707.05005
  • [27] Pan SR, 2018, PROCEEDINGS OF THE TWENTY-SEVENTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, P2609
  • [28] Symmetric Graph Convolutional Autoencoder for Unsupervised Graph Representation Learning
    Park, Jiwoong
    Lee, Minsik
    Chang, Hyung Jin
    Lee, Kyuewang
    Choi, Jin Young
    [J]. 2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2019), 2019, : 6518 - 6527
  • [29] Graph Representation Learning via Graphical Mutual Information Maximization
    Peng, Zhen
    Huang, Wenbing
    Luo, Minnan
    Zheng, Qinghua
    Rong, Yu
    Xu, Tingyang
    Huang, Junzhou
    [J]. WEB CONFERENCE 2020: PROCEEDINGS OF THE WORLD WIDE WEB CONFERENCE (WWW 2020), 2020, : 259 - 270
  • [30] GCC: Graph Contrastive Coding for Graph Neural Network Pre-Training
    Qiu, Jiezhong
    Chen, Qibin
    Dong, Yuxiao
    Zhang, Jing
    Yang, Hongxia
    Ding, Ming
    Wang, Kuansan
    Tang, Jie
    [J]. KDD '20: PROCEEDINGS OF THE 26TH ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY & DATA MINING, 2020, : 1150 - 1160