摘 要：为解决图神经网络（GNN）上不平衡节点的分类问题，提出一种 Bagging 集成模型，该模型使用图卷积网络（GCN）作为基分类器。在该模型中，先对若干基分类器进行并行训练，然后使用多数投票的方式对这些基分类器的预测结果进行集成，最终完成分类任务。实验结果表明，该文提出的模型显著优于其他现有基线方法，验证了其在不平衡节点分类中的有效性。
关键词：图神经网络；不平衡节点分类；集成学习
DOI:10.19850/j.cnki.2096-4706.2023.03.006
中图分类号：TP183 文献标识码：A 文章编号：2096-4706（2023）03-0029-04
Ensemble Graph Neural Network Model for Imbalanced Node Classification
GUO Mengxin
(Taiyuan Normal University, Jinzhong 030619, China)
Abstract: To solve the classification problem of unbalanced nodes on graph neural network (GNN), a Bagging ensemble model is proposed, which uses GCN as the base classifier. In this model, several base classifiers are trained in parallel, and then the prediction results of these base classifiers are integrated by majority voting to complete the classification task finally. Experimental results show that the proposed model in this paper is significantly superior to other existing baseline methods, and its effectiveness in unbalanced node classification is verified.
Keywords: graph neural network; imbalanced node classification; ensemble learning
参考文献：
[1] KIPF T N，WELLING M. Semi-Supervised Classification with Graph Convolutional Networks [J/OL].arXiv: 1609.02907 [cs.LG]. [2022-09-26].https://arxiv.org/abs/1609.02907v3.
[2] HAMILTON W L，YING R，LESKOVEC J. Inductive representation learning on large graphs [C]//NIPS'17：Proceedings of the 31st International Conference on Neural Information Processing Systems.Long Beach：Curran Associates Inc.，2017：1025–1035.
[3] SHI M，TANG Y F，ZHU X Q，et al. Multi-Class Imbalanced Graph Convolutional Network Learning [C]//Proceedings of the 29th International Joint Conference on Artificial Intelligence. Yokohama：[s.n.]，2021：2862-2868.
[4] ZHAO T X，ZHANG X，WANG S H. GraphSMOTE： Imbalanced Node Classification on Graphs with Graph Neural Networks [J/OL]. arXiv: 2103.08826 [cs.LG].[2022-09-20].https://arxiv.org/abs/2103.08826.
[5] BREIMAN L. Bagging Predictors [J].Machine learning， 1996，24（2）：123-140.
[6] FERNÁNDEZ A，GARCIA S，HERRERA F，et al. SMOTE for Learning from Imbalanced Data：Progress and Challenges，Marking the 15-year Anniversary [J].The Journal of Artificial Intelligence Research，2018，61：863-905.
[7] CHEN D L，LIN Y K，ZHAO G X，et al.Topology-Imbalance Learning for Semi-Supervised Node Classification [J/OL].arXiv: 2110.04099 [cs.LG].[2022-09-20].https://arxiv.org/abs/2110.04099.
[8] WU L R，LIN H T，GAO Z Y，et al. GraphMixup： Improving Class-Imbalanced Node Classification on Graphs by Selfsupervised Context Prediction [J/OL].arXiv: 2106.11133 [cs.LG].[2022- 09-20].https://arxiv.org/abs/2106.11133.
[9] WANG Y，AGGARWAL C，DERR T.Distance-wise Prototypical Graph Neural Network in Node Imbalance Classification [J/OL]. arXiv: 2110.12035 [cs.LG].[2022-09-20].https://arxiv.org/abs/2110.12035v1.
[10] LIU Y，AO X，QIN Z D，et al. Pick and Choose: A GNN-based Imbalanced Learning Approach for Fraud Detection [C]// Proceedings of the Web Conference 2021.Ljubljana：Association for Computing Machinery.2021：3168-3177.
[11] WANG Z，YE X J，WANG C K，et al. Network Embedding With Completely-Imbalanced Labels [J].IEEE Transactions on Knowledge and Data Engineering，2021，33（11）：3634-3647.
作者简介：郭梦昕（1996—），女，汉族，山西吕梁人，硕士研究生在读，研究方向：智能数据开发与应用。