基于解耦标签传播和多节点混合正则的图神经网络分类器

doi:10.16451/j.cnki.issn1003-6059.202505007

摘要
图/表
参考文献
相关文章 (15)

全文: PDF (797 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要图神经教师网络指导下的多层感知器在一定程度上平衡图数据相关任务中的推理性能与推理效率,但多层感知分类器独立看待图节点,难以显式捕获目标节点的邻域信息,推理性能受限.为此,文中提出基于解耦标签传播和多节点混合正则的图神经网分类器(Graph Neural Network Classifier Based on Decoupled Label Propagation and Multi-node Mixup Regularization, DLPMMR),基于知识蒸馏框架训练多层感知分类器,保证高推理效率下的基础推理性能.在训练阶段,基于朴素无超参数的二次组合策略实现多节点混合,增强节点多样性,并据此构建混合正则项,显式调控多层感知分类器的复杂性,提升其泛化性与鲁棒性.在推理阶段,引入标签传播,为多层感知分类器的推理纳入其所缺失的目标节点邻域信息,并解耦目标节点与邻域节点,有效控制邻居节点信息对目标节点分类决策的影响程度,进一步提升多层感知分类器的推理精度.在5个图节点分类基准数据集上的实验表明,DLPMMR自然鲁棒、性能较优.

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	贺文武
	刘小雨
	毛国君

关键词 ：图神经网络, 多层感知分类器, 知识蒸馏, 标签传播, 数据增强

Abstract：Graph neural network-distilled multilayer perceptrons(MLPs) balance inference performance and efficiency in graph-related tasks to some extent. However, MLPs treat graph nodes independently and struggle to explicitly capture neighborhood information of target nodes. Thus, their inference performance is limited. To solve this problem,a graph neural network classifier based on decoupled label propagation and multi-node mixup regularization(DLPMMR) is proposed. DLPMMR trains the MLP classifier under a knowledge distillation framework to ensure basic inference performance with high inference efficiency. During the training phase, a naive and hyperparameter-free double combination strategy is employed for multi-node mixup to enhance node diversity. A mixup regularization term is then constructed to explicitly control the complexity of the MLP so as to improve its generalization ability and robustness. During the inference phase, label propagation is introduced to incorporate missing neighborhood information into the predictions of the MLP. By decoupling target nodes from their neighboring nodes, the influence of neighbor node information on the classification decision of the target node is effectively regulated, and thus the inference accuracy of MLP is further enhanced. Experiments on five benchmark graph node classification datasets demonstrate that DLPMMR exhibits strong robustness and superior performance.

Key words： Graph Neural Network Multilayer Perceptron Classifier Knowledge Distillation Label Propagation Data Augmentation

收稿日期: 2025-04-07

ZTFLH:

TP391

基金资助:国家自然科学基金项目(No.41971340)、福建省自然科学基金项目(No.2024J01158)资助

通讯作者: 贺文武,博士,教授,主要研究方向为可信赖人工智能、图神经网络.E-mail: hwwhbb@163.com.

作者简介: 刘小雨,硕士研究生,主要研究方向为人工智能、图神经网络.E-mail:lxiaoyu0321@163.com.
毛国君,博士,教授,主要研究方向为人工智能、数据挖掘、分布式计算.E-mail: 19662092@fjut.edu.cn.

引用本文:

贺文武, 刘小雨, 毛国君. 基于解耦标签传播和多节点混合正则的图神经网络分类器[J]. 模式识别与人工智能, 2025, 38(5): 472-483. HE Wenwu, LIU Xiaoyu, MAO Guojun. Graph Neural Network Classifier Based onDecoupled Label Propagation and Multi-node Mixup Regularization. Pattern Recognition and Artificial Intelligence, 2025, 38(5): 472-483.

链接本文:

http://manu46.magtech.com.cn/Jweb_prai/CN/10.16451/j.cnki.issn1003-6059.202505007 或 http://manu46.magtech.com.cn/Jweb_prai/CN/Y2025/V38/I5/472

[1] GAO C, ZHENG Y, LI N, et al. A Survey of Graph Neural Networks for Recommender Systems: Challenges, Methods, and Directions. ACM Transactions on Recommender Systems, 2023, 1(1). DOI: 10.1145/356802.
[2] PURIFICATO E, BORATTO L, DE LUCA E W. Leveraging Graph Neural Networks for User Profiling: Recent Advances and Open Challenges // Proc of the 32nd ACM International Conference on Information and Knowledge Management. New York, USA: ACM, 2023: 5216-5219.
[3] LI Q, PENG H, LI J X, et al. A Survey on Text Classification: From Traditional to Deep Learning. ACM Transactions on Intelligent Systems and Technology, 2022, 13(2). DOI: 10.1145/34951.
[4] GOU J P, YU B S, MAYBANK S J, et al. Knowledge Distillation: A Survey. International Journal of Computer Vision, 2021, 129(6): 1789-1819.
[5] CHEN L, CHEN Z D, BRUNA J.On Graph Neural Networks Versus Graph-Augmented MLPS[C/OL]. [2025-03-17].https://arxiv.org/pdf/2010.15116.
[6] ZHANG S C, LIU Y Z, SUN Y Z, et al. Graph-Less Neural Networks: Teaching Old MLPS New Tricks via Distillation[C/OL].[2025-03-17]. https://arxiv.org/pdf/2110.08727.
[7] ZHENG W Q, HUANG E W, RAO N, et al. Cold Brew: Distilling Graph Node Representations with Incomplete or Missing Neighborhoods[C/OL].[2025-03-17]. https://arxiv.org/pdf/2111.04840.
[8] ZHANG H Y, CISSE M, DAUPHIN Y N, et al. mixup: Beyond Empirical Risk Minimization[C/OL].[2025-03-17]. https://arxiv.org/pdf/1710.09412.pdf.
[9] VERMA V, QU M, KAWAGUCHI K, et al. GraphMix: Improved Training of GNNS for Semi-Supervised Learning. Proceedings of the AAAI Conference on Artificial Intelligence, 2021, 35(11): 10024-10032.
[10] LU W G, GUAN Z Y, ZHAO W, et al. NodeMixup: Tackling Under-Reaching for Graph Neural Networks. Proceedings of the AAAI Conference on Artificial Intelligence, 2024, 38(13): 14175-14183.
[11] KIPF T N, WELLING M.Semi-supervised Classification with Graph Convolutional Networks[C/OL]. [2025-03-17].https://arxiv.org/pdf/1609.02907.
[12] VELIČKOVIĆ P, CUCURULL G, CASANOVA A, et al. Graph Attention Networks[C/OL].[2025-03-17]. https://arxiv.org/pdf/1710.10903.
[13] HAMILTON W L, YING R, LESKOVEC J.Inductive Representation Learning on Large Graphs // Proc of the 31st International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2017: 1025-1035.
[14] CHEN J, MA T F, XIAO C.FastGCN: Fast Learning with Graph Convolutional Networks via Importance Sampling[C/OL]. [2025-03-17].https://arxiv.org/pdf/1801.10247.
[15] YANG Y D, QIU J Y, SONG M L, et al. Distilling Knowledge from Graph Convolutional Networks // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2020: 7072-7081.
[16] YAN B C, WANG C K, GUO G Y, et al. TinyGNN: Learning Efficient Graph Neural Networks // Proc of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York, USA: ACM, 2020: 1848-1856.
[17] LU W G, GUAN Z Y, ZHAO W, et al. AdaGMLP: AdaBoosting GNN-to-MLP Knowledge Distillation // Proc of the 30th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York, USA: ACM, 2024: 2060-2071.
[18] TIAN Y J, ZHANG C X, GUO Z C, et al. Learning MLPs on Graphs: A Unified View of Effectiveness, Robustness, and Efficiency[C/OL].[2025-03-17]. https://openreview.net/pdf?id=Cs3r5KLdoj.
[19] KIM J, YUN S, PARK C.S-Mixup: Structural Mixup for Graph Neural Networks // Proc of the 32nd ACM International Conference on Information and Knowledge Management. New York, USA: ACM, 2023: 4003-4007.
[20] BRAHIM L, LOUBNA B, ALI I.A Literature Survey on Label Pro-pagation for Community Detection // Proc of the 5th International Conference on Intelligent Computing in Data Sciences. Washington, USA: IEEE, 2021. DOI: 10.1109/ICDS53782.2021.9626716.
[21] GHAYEKHLOO M, NICKABADI A.CLP-GCN: Confidence and La-bel Propagation Applied to Graph Convolutional Networks. Applied Soft Computing, 2023, 132. DOI: 10.1016/j.asoc.2022.109850.
[22] WANG H W, LESKOVEC J.Unifying Graph Convolutional Neural Networks and Label Propagation[C/OL]. [2025-03-17].https://arxiv.org/pdf/2002.06755.
[23] YANG C, LIU J W, SHI C.Extract the Knowledge of Graph Neural Networks and Go Beyond It: An Effective Knowledge Distillation Framework // Proc of the Web Conference. New York, USA: ACM, 2021:1227-1237.
[24] XU H, XIANG L Y, HUANG F, et al. GRACE: Graph Self-Distillation and Completion to Mitigate Degree-Related Biases // Proc of the 29th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York, USA: ACM, 2023: 2813-2824.
[25] SEN P, NAMATA G, BILGIC M, et al. Collective Classification in Network Data. AI Magazine, 2008, 29(3): 93-106.
[26] NAMATA G, LONDON B, GETOOR L, et al. Query-Driven Active Surveying for Collective Classification[C/OL].[2025-03-17]. https://dtai.cs.kuleuven.be/events/mlg2012/papers/11_querying_namata.pdf.
[27] SHCHUR O, MUMME M, BOJCHEVSKI A, et al. Pitfalls of Graph Neural Network Evaluation[C/OL].[2025-03-17]. https://arxiv.org/pdf/1811.05868.
[28] DU H Y, YU R, BAI L, et al. Learning Structure Perception MLPs on Graphs: A Layer-Wise Graph Knowledge Distillation Framework. International Journal of Machine Learning and Cybernetics, 2024, 15(10): 4357-4372.