基于双重对比学习和硬样本挖掘的多视图图聚类算法

doi:10.19678/j.issn.1000-3428.0069632

摘要/Abstract

摘要：

图聚类作为图挖掘领域的关键研究方向，旨在从图数据中发现具有相似性的子结构或节点群体，将它们划分到同一簇。多视图图聚类算法通过综合图数据的多个视图，充分利用底层信息以提升聚类质量。近年来，图对比学习的改进推动了基于深度图学习的多视图图聚类的迅速发展。然而，现有的图对比学习难以提高节点表示的识别性。针对上述问题，提出一种基于双重对比学习和硬样本挖掘的多视图图聚类方法。首先，通过图滤波器平滑节点表示以减轻噪声节点带来的影响；然后，设计节点紧凑性对比学习和节点一致性对比学习，以提高同一聚类内节点表示的紧凑性和不同视图间节点表示的一致性；最后，考虑到基于图对比学习的多视图图聚类存在假阴性问题，提出基于聚类引导的硬样本挖掘策略，以提高多视图图聚类效果。在ACM、DBLP和IMDB 3个真实世界数据集上进行实验，结果表明，该方法分别取得了94.49%、93.22%和57.51%的准确率，均高于8种对比基线方法。

关键词: 图挖掘, 多视图, 图聚类, 深度图学习, 图对比学习

Abstract:

As a key research direction in the field of graph mining, graph clustering aims to discover substructures or node groups with similarities from graph data and classify them into the same cluster. The multi-view graph clustering algorithm integrates multiple views of graph data and fully utilizes the underlying information to improve the clustering quality. In recent years, improvements in graph contrastive learning have driven the rapid development of multi-view graph clustering based on deep graph learning. However, improving the recognition of node representations using existing graph contrastive learning methods is challenging. A multi-view graph clustering method based on dual contrastive learning and hard sample mining is proposed to address these issues. First, the node representation is smoothed using graph filters to mitigate the impact of noisy nodes. Subsequently, node compactness comparison learning and node consistency comparison learning are designed to improve the compactness of node representations within the same cluster and the consistency of node representations between different views. Finally, considering the issue of false negatives in multi-view graph clustering based on graph contrastive learning, a hard sample mining strategy guided by clustering is proposed to improve the clustering performance of multi-view graphs. Experiments conducted on three real-world datasets, ACM, DBLP, and IMDB, reveal that the proposed method achieves accuracies of 94.49%, 93.22%, and 57.51%, respectively. These values are higher than those of the eight baseline comparison methods.

Key words: graph mining, multi-view, graph clustering, deep graph learning, graph contrastive learning

钱李烽, 李静, 邹徐熹, 陈宇, 顾亚林, 魏训虎. 基于双重对比学习和硬样本挖掘的多视图图聚类算法[J]. 计算机工程, 2025, 51(12): 82-95.

QIAN Lifeng, LI Jing, ZOU Xuxi, CHEN Yu, GU Yalin, WEI Xunhu. Multi-view Graph Clustering Algorithm Based on Dual Contrastive Learning and Hard Sample Mining[J]. Computer Engineering, 2025, 51(12): 82-95.

https://www.ecice06.com/CN/Y2025/V51/I12/82

图/表 14

图1 MVC-DCLHSM整体框架

Fig.1 Overall framework of MVC-DCLHSM

图2 消融分析结果

Fig.2 Ablation analysis results

图3 数据集ACM上的参数敏感性分析结果1

Fig.3 Parameter sensitivity analysis results 1 on ACM dataset

图4 数据集DBLP上的参数敏感性分析结果1

Fig.4 Parameter sensitivity analysis results 1 on DBLP dataset

图5 数据集IMDB上的参数敏感性分析结果1

Fig.5 Parameter sensitivity analysis results 1 on IMDB dataset

图6 数据集ACM上的参数敏感性分析结果2

Fig.6 Parameter sensitivity analysis results 2 on ACM dataset

图7 数据集DBLP上的参数敏感性分析结果2

Fig.7 Parameter sensitivity analysis results 2 on DBLP dataset

图8 数据集IMDB上的参数敏感性分析结果2

Fig.8 Parameter sensitivity analysis results 2 on IMDB dataset

图9 数据集ACM上的可视化分析

Fig.9 Visualization and analysis of ACM dataset

图10 数据集DBLP上的可视化分析

Fig.10 Visualization and analysis of DBLP dataset

图11 数据集IMDB上的可视化分析

Fig.11 Visualization and analysis of IMDB dataset

参考文献 35

1	VELIČKOVIĆ P . Everything is connected: graph neural networks. Current Opinion in Structural Biology, 2023, 79, 102538. doi: 10.1016/j.sbi.2023.102538
2	YANG X H , HU X C , ZHOU S H , et al. Interpolation-based contrastive learning for few-label semi-supervised learning. IEEE Transactions on Neural Networks and Learning Systems, 2024, 35 (2): 2054- 2065. doi: 10.1109/TNNLS.2022.3186512
3	LIU H , WANG T , LI Y D , et al. Joint graph learning and matching for semantic feature correspondence. Pattern Recognition, 2023, 134, 109059. doi: 10.1016/j.patcog.2022.109059
4	XU J, TANG H Y, REN Y Z, et al. Multi-level feature learning for contrastive multi-view clustering[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D.C., USA: IEEE Press, 2022: 16030-16039.
5	LIN R J , LIN Y K , LIN Z H , et al. CCR-Net: consistent contrastive representation network for multi-view clustering. Information Sciences, 2023, 637, 118937. doi: 10.1016/j.ins.2023.118937
6	刘鹏仪, 胡节, 王红军, 等. 基于对比共识图学习的多视图属性图聚类算法. 计算机科学, 2024, 51 (11): 73- 80.
	LIU P Y , HU J , WANG H J , et al. A multi-view attribute graph clustering algorithm based on comparative consensus graph learning. Computer Science, 2024, 51 (11): 73- 80.
7	LIN Z, KANG Z. Graph filter-based multi-view attributed graph clustering[EB/OL]. [2024-01-05]. https://www.ijcai.org/proceedings/2021/0375.pdf.
8	PAN E , KANG Z . Multi-view contrastive graph clustering. Advances in Neural Information Processing Systems, 2021, 34 (1): 2148- 2159.
9	ZENG J Q , XIE P T . Contrastive self-supervised learning for graph classification. Proceedings of the AAAI Conference on Artificial Intelligence, 2021, 35 (12): 10824- 10832. doi: 10.1609/aaai.v35i12.17293
10	WAN S , PAN S R , YANG J , et al. Contrastive and generative graph convolutional networks for graph-based semi-supervised learning. Proceedings of the AAAI Conference on Artificial Intelligence, 2021, 35 (11): 10049- 10057. doi: 10.1609/aaai.v35i11.17206
11	DONG W, WU J S, LUO Y, et al. Node representation learning in graph via node-to-neighbourhood mutual information maximization[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Washington D.C., USA: IEEE Press, 2022: 16599-16608.
12	SOUZA F , PREMEBIDA C , ARAÚJO R . High-order conditional mutual information maximization for dealing with high-order dependencies in feature selection. Pattern Recognition, 2022, 131, 108895. doi: 10.1016/j.patcog.2022.108895
13	YANG X, LIU Y, ZHOU S, et al. Cluster-guided contrastive graph clustering network[EB/OL]. [2024-01-05]. https://arxiv.org/abs/2301.01098.
14	MO S, SUN Z, LI C. Siamese prototypical contrastive learning[EB/OL]. [2024-01-05]. https://arxiv.org/abs/2208.08819.
15	FANG U , LI M , LI J X , et al. A comprehensive survey on multi-view clustering. IEEE Transactions on Knowledge and Data Engineering, 2023, 35 (12): 12350- 12368. doi: 10.1109/TKDE.2023.3270311
16	THAKOOR S, TALLEC C, AZAR M G, et al. Bootstrapped representation learning on graphs[EB/OL]. [2024-01-05]. https://openreview.net/forum?id=QrzVRAA49Ud.
17	WANG H , YANG Y , LIU B . GMC: graph-based multi-view clustering. IEEE Transactions on Knowledge and Data Engineering, 2020, 32 (6): 1116- 1129. doi: 10.1109/TKDE.2019.2903810
18	LIU Y , YANG X H , ZHOU S H , et al. Hard sample aware network for contrastive deep graph clustering. Proceedings of the AAAI Conference on Artificial Intelligence, 2023, 37 (7): 8914- 8922. doi: 10.1609/aaai.v37i7.26071
19	ZHOU J , CUI G Q , HU S D , et al. Graph neural networks: a review of methods and applications. AI Open, 2020, 1, 57- 81. doi: 10.1016/j.aiopen.2021.01.001
20	赵振廷, 赵旭俊. 多样性约束和高阶信息挖掘的多视图聚类. 计算机应用研究, 2024, 41 (8): 2309- 2314.
	ZHAO Z T , ZHAO X J . Multi-view clustering with diversity constraints and higher-order information mining. Computer Applications Research, 2024, 41 (8): 2309- 2314.
21	YANG X H, JIN J Q, WANG S W, et al. DealMVC: dual contrastive calibration for multi-view clustering[C]//Proceedings of the 31st ACM International Conference on Multimedia. New York, USA: ACM Press, 2023: 337-346.
22	朱云华, 孔兵, 周丽华, 等. 图对比学习引导的多视图聚类网络. 计算机应用, 2024, 44 (10): 3267- 3274.
	ZHU Y H , KONG B , ZHOU L H , et al. Figure comparison of learning-guided multi-view clustering networks. Computer Applications, 2024, 44 (10): 3267- 3274.
23	CUI G Q, ZHOU J, YANG C, et al. Adaptive graph encoder for attributed graph embedding[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. New York, USA: ACM Press, 2020: 976-985.
24	CHUNG F R K . Spectral graph theory. [S.l.]: American Mathematical Soc., 1997.
25	WANG C, PAN S, HU R, et al. Attributed graph clustering: a deep attentional embedding approach[EB/OL]. [2024-01-05]. https://arxiv.org/abs/1906.06532.
26	LÓPEZ-ORIONA Á , VILAR J A , D'URSO P . Hard and soft clustering of categorical time series based on two novel distances with an application to biological sequences. Information Sciences, 2023, 624, 467- 492. doi: 10.1016/j.ins.2022.12.065
27	CHEN M S , LIN J Q , LI X L , et al. Representation learning in multi-view clustering: a literature review. Data Science and Engineering, 2022, 7 (3): 225- 241. doi: 10.1007/s41019-022-00190-8
28	TANG J, QU M, WANG M Z, et al. LINE: large-scale information network embedding[EB/OL]. [2024-01-05]. https://arxiv.org/abs/1503.03578.
29	KIPF T N, WELLING M. Variational graph auto-encoders[EB/OL]. [2024-01-05]. http://bayesiandeeplearning.org/2016/papers/BDL_16.pdf.
30	LIU W Y, CHEN P Y, YEUNG S, et al. Principled multilayer network embedding[C]//Proceedings of the IEEE International Conference on Data Mining Workshops (ICDMW). Washington D.C., USA: IEEE Press, 2017: 134-141.
31	XIA R K , PAN Y , DU L , et al. Robust multi-view spectral clustering via low-rank and sparse decomposition. Proceedings of the AAAI Conference on Artificial Intelligence, 2014, 28 (1): 1- 7.
32	FAN S H, WANG X, SHI C, et al. One2Multi graph autoencoder for multi-view graph clustering[EB/OL].[2024-01-05]. https://dl.acm.org/doi/abs/10.1145/3366423.3380079.
33	YI S Y , JU W , QIN Y F , et al. Redundancy-free self-supervised relational learning for graph clustering. IEEE Transactions on Neural Networks and Learning Systems, 2024, 35 (12): 18313- 18327. doi: 10.1109/TNNLS.2023.3314451
34	ZHANG M J , WANG D W , WU H R , et al. Multi-view contrastive learning for multilayer network embedding. Journal of Computational Science, 2023, 67, 101975. doi: 10.1016/j.jocs.2023.101975
35	BO D Y, WANG X, SHI C, et al. Structural deep clustering network[EB/OL].[2024-01-05]. https://arxiv.org/abs/2002.01633.

[1]	林合川, 徐慧英, 朱信忠, 黄晓, 刘子洋. 自加权多视图k-均值算法[J]. 计算机工程, 2025, 51(8): 141-150.
[2]	雷一凡, 陈晓红. 隐私保护的去中心联邦多视图聚类[J]. 计算机工程, 2025, 51(7): 180-189.
[3]	姚迅, 王海鹏, 胡新荣, 杨捷. 基于自适应增强的多视图对比推荐算法[J]. 计算机工程, 2025, 51(5): 103-113.
[4]	杨春霞, 蒋耀, 翟雪彤, 周媛媛. 基于一致性功能神经过程的多视图时序预测[J]. 计算机工程, 2025, 51(10): 111-120.
[5]	胡傲然, 陈晓红. 基于多样性与一致性的单步多视图聚类[J]. 计算机工程, 2024, 50(5): 51-61.
[6]	谢欢, 刘纯平, 季怡. 基于单-多视图优化的足球球员三维姿态和体型估计[J]. 计算机工程, 2024, 50(3): 200-207.
[7]	杜晓英, 袁庆霓, 齐建友, 王晨, 杜飞龙, 任澳. 动态场景下基于语义分割的视觉SLAM方法[J]. 计算机工程, 2024, 50(3): 242-249.
[8]	袁文涛, 卫文韬, 高德民. 融合注意力机制的多视图卷积手势识别研究[J]. 计算机工程, 2024, 50(3): 208-215.
[9]	王思成, 江浩, 陈晓. 具有跨尺度Transformer的高效多视图立体网络[J]. 计算机工程, 2024, 50(11): 266-275.
[10]	顾嘉静, 杨丹, 聂铁铮, 寇月. 基于多视图融合跨层对比学习的推荐算法[J]. 计算机工程, 2024, 50(1): 120-128.
[11]	刘思慧, 高全学, 宋伟, 谢德燕. 基于加权张量低秩约束的多视图谱聚类[J]. 计算机工程, 2024, 50(1): 129-137.
[12]	赵晓杰, 牛雪莹, 张继福. 基于联合熵的多视图集成聚类分析[J]. 计算机工程, 2023, 49(10): 112-119.
[13]	王曙燕, 郭睿涵, 孙家泽. 基于图对比学习的MOOC推荐方法[J]. 计算机工程, 2023, 49(1): 57-64,72.
[14]	贺娜, 马盈仓. 融合KL信息的多视图模糊聚类算法[J]. 计算机工程, 2022, 48(7): 114-121,150.
[15]	房立金, 王科棋. 一种结合深度学习的运动重检测视觉SLAM算法[J]. 计算机工程, 2022, 48(5): 18-26.

选择文件类型/文献管理软件名称

选择包含的内容