《计算机应用》唯一官方网站 ›› 2026, Vol. 46 ›› Issue (5): 1370-1377.DOI: 10.11772/j.issn.1001-9081.2025050610
• 人工智能 • 上一篇
收稿日期:2025-06-04
修回日期:2025-10-11
接受日期:2025-10-14
发布日期:2025-10-29
出版日期:2026-05-10
通讯作者:
贺超波
作者简介:郑宝源(2003—),男,广东揭阳人,硕士研究生,主要研究方向:图数据挖掘、图神经网络
基金资助:Received:2025-06-04
Revised:2025-10-11
Accepted:2025-10-14
Online:2025-10-29
Published:2026-05-10
Contact:
Chaobo HE
About author:ZHENG Baoyuan, born in 2003, M. S. candidate. His research interests include graph data mining, graph neural networks.
Supported by:摘要:
图卷积网络(GCN)在图表示学习领域已展现了强大的潜力,然而,现有的GCN方法在全局拓扑关系学习以及拓扑结构和属性特征融合方面仍存在局限性。针对该问题,提出一种图扩散与双视图特征学习增强的图卷积网络(GCN-GDDV)。该网络首先引入广义图扩散机制构建包含全局拓扑结构信息的扩散图;随后,结合属性特征K近邻图进行基于GCN的双视图特征学习,以分别捕捉全局结构关系依赖与节点属性的语义相似性;最后,设计注意力网络实现拓扑结构和属性特征的自适应融合。在3个常用图数据集上进行节点分类实验的结果表明:GCN-GDDV相较于次优方法在准确率、Macro-F1和Micro-F1指标上分别提升1.78%、1.60%和0.30%。
中图分类号:
郑宝源, 贺超波. 图扩散与双视图特征学习增强的图卷积网络[J]. 计算机应用, 2026, 46(5): 1370-1377.
Baoyuan ZHENG, Chaobo HE. Graph convolutional network enhanced by graph diffusion and dual-view feature learning[J]. Journal of Computer Applications, 2026, 46(5): 1370-1377.
表1 数据集的统计信息
Tab. 1 Dataset statistics?
| 模型 | Cora | PubMed | CiteSeer | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 准确率 | Macro-F1 | Micro-F1 | 准确率 | Macro-F1 | Micro-F1 | 准确率 | Macro-F1 | Micro-F1 | |
| GCN | 0.801 0 | 0.789 0 | 0.793 0 | 0.773 0 | 0.765 0 | 0.762 0 | 0.721 0 | 0.719 0 | 0.707 0 |
| GAT | 0.803 0 | 0.793 0 | 0.788 0 | 0.762 0 | 0.758 0 | 0.755 0 | 0.692 0 | 0.688 0 | 0.684 0 |
| APPNP | 0.827 0 | 0.825 0 | 0.816 0 | 0.802 0 | 0.793 0 | 0.796 0 | 0.715 0 | 0.709 0 | 0.697 0 |
| GDC | 0.827 0 | 0.821 0 | 0.824 0 | 0.794 0 | 0.792 0 | 0.798 0 | 0.721 0 | 0.728 0 | |
| ADC | 0.825 0 | 0.819 0 | 0.788 0 | 0.770 0 | 0.776 0 | 0.705 0 | 0.716 0 | 0.709 0 | |
| AGMixup | 0.824 0 | 0.826 0 | 0.822 0 | 0.798 0 | 0.795 0 | 0.789 0 | 0.712 0 | 0.717 0 | 0.715 0 |
| NLGT | 0.825 0 | 0.822 0 | 0.801 0 | 0.796 0 | 0.798 0 | 0.714 0 | 0.708 0 | 0.712 0 | |
| HiD-Net | 0.831 0 | 0.821 0 | 0.792 0 | 0.717 0 | |||||
| GCN-GDDV | 0.843 0 | 0.839 0 | 0.833 0 | 0.814 0 | 0.808 0 | 0.742 0 | 0.739 0 | 0.728 0 | |
表2 不同方法的分类性能对比
Tab. 2 Classification performance comparison of different methods
| 模型 | Cora | PubMed | CiteSeer | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 准确率 | Macro-F1 | Micro-F1 | 准确率 | Macro-F1 | Micro-F1 | 准确率 | Macro-F1 | Micro-F1 | |
| GCN | 0.801 0 | 0.789 0 | 0.793 0 | 0.773 0 | 0.765 0 | 0.762 0 | 0.721 0 | 0.719 0 | 0.707 0 |
| GAT | 0.803 0 | 0.793 0 | 0.788 0 | 0.762 0 | 0.758 0 | 0.755 0 | 0.692 0 | 0.688 0 | 0.684 0 |
| APPNP | 0.827 0 | 0.825 0 | 0.816 0 | 0.802 0 | 0.793 0 | 0.796 0 | 0.715 0 | 0.709 0 | 0.697 0 |
| GDC | 0.827 0 | 0.821 0 | 0.824 0 | 0.794 0 | 0.792 0 | 0.798 0 | 0.721 0 | 0.728 0 | |
| ADC | 0.825 0 | 0.819 0 | 0.788 0 | 0.770 0 | 0.776 0 | 0.705 0 | 0.716 0 | 0.709 0 | |
| AGMixup | 0.824 0 | 0.826 0 | 0.822 0 | 0.798 0 | 0.795 0 | 0.789 0 | 0.712 0 | 0.717 0 | 0.715 0 |
| NLGT | 0.825 0 | 0.822 0 | 0.801 0 | 0.796 0 | 0.798 0 | 0.714 0 | 0.708 0 | 0.712 0 | |
| HiD-Net | 0.831 0 | 0.821 0 | 0.792 0 | 0.717 0 | |||||
| GCN-GDDV | 0.843 0 | 0.839 0 | 0.833 0 | 0.814 0 | 0.808 0 | 0.742 0 | 0.739 0 | 0.728 0 | |
| 数据集 | w/oDADM | w/oAM | GCN-GDDV | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 准确率 | Macro-F1 | Micro-F1 | 准确率 | Macro-F1 | Micro-F1 | 准确率 | Macro-F1 | Micro-F1 | |
| Cora | 0.822 0 | 0.816 0 | 0.809 0 | 0.825 0 | 0.820 0 | 0.814 0 | 0.843 0 | 0.839 0 | 0.833 0 |
| PubMed | 0.799 0 | 0.786 0 | 0.778 0 | 0.793 0 | 0.781 0 | 0.773 0 | 0.814 0 | 0.808 0 | 0.796 0 |
| CiteSeer | 0.712 0 | 0.693 0 | 0.681 0 | 0.697 0 | 0.685 0 | 0.672 0 | 0.742 0 | 0.739 0 | 0.728 0 |
表 3 消融实验结果
Tab. 3 Ablation study results?
| 数据集 | w/oDADM | w/oAM | GCN-GDDV | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 准确率 | Macro-F1 | Micro-F1 | 准确率 | Macro-F1 | Micro-F1 | 准确率 | Macro-F1 | Micro-F1 | |
| Cora | 0.822 0 | 0.816 0 | 0.809 0 | 0.825 0 | 0.820 0 | 0.814 0 | 0.843 0 | 0.839 0 | 0.833 0 |
| PubMed | 0.799 0 | 0.786 0 | 0.778 0 | 0.793 0 | 0.781 0 | 0.773 0 | 0.814 0 | 0.808 0 | 0.796 0 |
| CiteSeer | 0.712 0 | 0.693 0 | 0.681 0 | 0.697 0 | 0.685 0 | 0.672 0 | 0.742 0 | 0.739 0 | 0.728 0 |
| 数据集 | GCN | GAT | APPNP | GDC | ADC | AGMixup | NLGT | HiD-Net | GCN-GDDV |
|---|---|---|---|---|---|---|---|---|---|
| CiteSeer | 78.1 | 90.3 | 83.9 | 111.0 | 119.0 | 108.0 | 135 | 84.7 | 133 |
| PubMed | 63.8 | 78.2 | 75.5 | 90.9 | 97.3 | 94.3 | 109 | 70.5 | 104 |
表4 模型训练时间对比 ( s)
Tab. 4 Comparison of model training time?
| 数据集 | GCN | GAT | APPNP | GDC | ADC | AGMixup | NLGT | HiD-Net | GCN-GDDV |
|---|---|---|---|---|---|---|---|---|---|
| CiteSeer | 78.1 | 90.3 | 83.9 | 111.0 | 119.0 | 108.0 | 135 | 84.7 | 133 |
| PubMed | 63.8 | 78.2 | 75.5 | 90.9 | 97.3 | 94.3 | 109 | 70.5 | 104 |
| 节点编号 | 节点类别 | GCN | GAT | APPNP | GDC | ADC | AGMixup | NLGT | HiD-Net | GCN-GDDV |
|---|---|---|---|---|---|---|---|---|---|---|
| 1511 | PM | RL(×) | RL(×) | RL(×) | PM(√) | PM(√) | PM(√) | RL(×) | RL(×) | PM(√) |
| 2003 | PM | RL(×) | RL(×) | PM(√) | RL(×) | PM(√) | PM(×) | PM(√) | PM(√) | PM(√) |
| 2348 | PM | PM(√) | PM(√) | RL(×) | PM(√) | RL(×) | RL(×) | PM(√) | PM(√) | PM(√) |
表5 案例及预测结果对比
Tab. 5 Comparison of case studies and forecast results
| 节点编号 | 节点类别 | GCN | GAT | APPNP | GDC | ADC | AGMixup | NLGT | HiD-Net | GCN-GDDV |
|---|---|---|---|---|---|---|---|---|---|---|
| 1511 | PM | RL(×) | RL(×) | RL(×) | PM(√) | PM(√) | PM(√) | RL(×) | RL(×) | PM(√) |
| 2003 | PM | RL(×) | RL(×) | PM(√) | RL(×) | PM(√) | PM(×) | PM(√) | PM(√) | PM(√) |
| 2348 | PM | PM(√) | PM(√) | RL(×) | PM(√) | RL(×) | RL(×) | PM(√) | PM(√) | PM(√) |
| [1] | JIN M, KOH H Y, WEN Q, et al. A survey on graph neural networks for time series: forecasting, classification, imputation, and anomaly detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024, 46(12): 10466-10485. |
| [2] | ZHU Y, FENG L, DENG Z, et al. Robust node classification on graph data with graph and label noise[C]// Proceedings of the 38th AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2024: 17220-17227. |
| [3] | 李鑫,陆伟,马召祎,等. 基于图注意力和改进Transformer的节点分类方法[J]. 电子学报, 2024, 52(8): 2799-2810. |
| LI X, LU W, MA Z Y, et al. A node classification method based on graph attention and improved Transformer[J]. Acta Electronica Sinica, 2024, 52(8): 2799-2810. | |
| [4] | ZHAO T, ZHANG X, WANG S. Disambiguated node classification with graph neural networks[C]// Proceedings of the ACM Web Conference 2024. New York: ACM, 2024: 914-923. |
| [5] | NGUYEN T K, FANG Y. Diffusion-based negative sampling on graphs for link prediction[C]// Proceedings of the ACM Web Conference 2024, New York: ACM, 2024: 948-958. |
| [6] | KUMAR M, MISHRA S, SINGH S S, et al. Community-enhanced link prediction in dynamic networks[J]. ACM Transactions on the Web, 2024, 18(2): No.24. |
| [7] | YANG L, WU L, ZHANG C, et al. Research on node classification and link prediction of knowledge graph in complex network analysis[C]// Proceedings of the IEEE 2nd International Conference on Electrical, Automation and Computer Engineering. Piscataway: IEEE, 2024: 1691-1695. |
| [8] | 刘群,谭洪胜,张优敏,等.基于元路径的动态异质网络表示学习[J].电子学报,2022,50(8):1830-1839. |
| LIU Q, TAN H S, ZHANG Y M, et al. Dynamic heterogeneous network representation method based on meta-path[J]. Acta Electronica Sinica, 2022, 50(8): 1830-1839. | |
| [9] | KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks[EB/OL]. [2024-12-13].. |
| [10] | 徐冰冰,岑科廷,黄俊杰,等.图卷积神经网络综述[J].计算机学报,2020,43(5):755-780. |
| XU B B, CEN K T, HUANG J J, et al. A survey on graph convolutional neural network[J]. Chinese Journal of Computers, 2020, 43(5): 755-780. | |
| [11] | REN W, JIN N, OUYANG L. Phase space graph convolutional network for chaotic time series learning[J]. IEEE Transactions on Industrial Informatics, 2024, 20(5): 7576-7584. |
| [12] | VELIČKOVIĆ P, CUCURULL G, CASANOVA A, et al. Graph attention networks[EB/OL]. [2025-03-11].. |
| [13] | HAMILTON W L, YING R, LESKOVEC J. Inductive representation learning on large graphs[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2017: 1025-1035. |
| [14] | ABU-EL-HAIJA S, PEROZZI B, KAPOOR A, et al. MixHop: higher-order graph convolutional architectures via sparsified neighborhood mixing[C]// Proceedings of the 36th International Conference on Machine Learning. New York: JMLR.org, 2019: 21-29. |
| [15] | GASTEIGER J, WEIßENBERGER S, GÜNNEMANN S. Diffusion improves graph learning[C]// Proceedings of the 33rd International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc, 2019: 13366-13378. |
| [16] | XU X, WANG T, YANG Y, et al. Radial graph convolutional network for visual question generation[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 32(4): 1654-1667. |
| [17] | TAILOR S A, OPOLKA F L, LIÒ P, et al. Do we need isotropic graph neural networks?[EB/OL]. [2025-04-06].. |
| [18] | LIU K, GAO L, KHAN N M, et al. A multi-stream graph convolutional networks-hidden conditional random field model for skeleton-based action recognition[J]. IEEE Transactions on Multimedia, 2021, 23: 64-76. |
| [19] | 郭新超,张维玉,夏忠秀.双向数据增强图卷积网络[J].计算机工程与设计,2023,44(8):2345-2351. |
| GUO X C, ZHANG W Y, XIA Z X. Two-way data augmentation graph convolutional networks[J]. Computer Engineering and Design, 2023, 44(8): 2345-2351. | |
| [20] | 郭向星,周魏,杨正益,等. 基于自监督图卷积和注意力机制实现隐式反馈降噪的社交推荐[J]. 电子学报, 2025, 53(1): 151-162. |
| GUO X X, ZHOU W, YANG Z Y, et al. Denoising implicit feedback with self-supervised graph convolution network and attention mechanism for social recommendation[J]. Acta Electronica Sinica, 2025, 53(1): 151-162. | |
| [21] | HUANG J, DU L, CHEN X, et al. Robust mid-pass filtering graph convolutional networks[C]// Proceedings of the ACM Web Conference 2023. New York: ACM, 2023: 328-338. |
| [22] | 周俊,胡斌斌,张志强,等. MoGE:基于图上下文增强的多任务推荐算法[J]. 电子学报, 2023, 51(11): 3377-3387. |
| ZHOU J, HU B B, ZHANG Z Q, et al. MoGE: graph context enhanced multi-task recommendation method[J]. Acta Electronica Sinica, 2023, 51(11): 3377-3387. | |
| [23] | GASTEIGER J, BOJCHEVSKI A, GÜNNEMANN S. Predict then propagate: graph neural networks meet personalized PageRank[EB/OL]. [2025-02-01].. |
| [24] | CHAMBERLAIN B P, ROWBOTTOM J, GORINOVA M, et al. GRAND: graph neural diffusion[C]// Proceedings of the 38th International Conference on Machine Learning. New York: JMLR.org, 2021: 1407-1418. |
| [25] | LI Y, WANG X, LIU H, et al. A generalized neural diffusion framework on graphs[C]// Proceedings of the 38th AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2024: 8707-8715. |
| [26] | MA Y, ZHAN K. Self-contrastive graph diffusion network[C]// Proceedings of the 31st ACM International Conference on Multimedia. New York: ACM, 2023: 3857-3865. |
| [27] | ZHAO J, DONG Y, DING M, et al. Adaptive diffusion in graph neural networks[C]// Proceedings of the 35th International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2021: 23321-23333. |
| [28] | LU W, GUAN Z, ZHAO W, et al. AGMixup: adaptive graph mixup for semi-supervised node classification[C]// Proceedings of the 39th AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2025: 19143-19151. |
| [29] | XU X, ZHOU Y, XIANG H, et al. NLGT: neighborhood-based and label-enhanced graph Transformer framework for node classification[C]// Proceedings of the 39th AAAI Conference on Artificial Intelligence. Palo Alto: AAAI Press, 2025: 12954-12962. |
| [1] | 白翔, 李巨川, 王慧民, 景超, 钮键, 张兴忠, 程永强. 基于改进Swin Transformer的电力图像检索方法[J]. 《计算机应用》唯一官方网站, 2026, 46(4): 1334-1343. |
| [2] | 秦传东, 索志强. 融合改进的ResNet50与集成分类器的皮肤癌分类[J]. 《计算机应用》唯一官方网站, 2026, 46(4): 1354-1362. |
| [3] | 豆旭梦, 解滨, 张朝晖, 赵振刚, 段菡煜, 郭澳磊. 基于结构‒网络协同特征与网格注意力增强KAN的药物靶标相互作用预测[J]. 《计算机应用》唯一官方网站, 2026, 46(4): 1344-1353. |
| [4] | 李文浩, 郭银章. 基于双层多尺度动态GCN模型的城市交通流量预测[J]. 《计算机应用》唯一官方网站, 2026, 46(4): 1323-1333. |
| [5] | 刘欢娴, 王洪涛, 王宪奥, 王洪梅, 徐伟峰. 跨模态语义关联的多模态事实验证[J]. 《计算机应用》唯一官方网站, 2026, 46(4): 1069-1076. |
| [6] | 邵培荣, 蔺素珍, 王彦博. 以人为中心的细节增强虚拟试衣方法[J]. 《计算机应用》唯一官方网站, 2026, 46(3): 915-923. |
| [7] | 张祖习, 张战成, 胡伏原. 局部与长程时序互补建模的视频动作识别[J]. 《计算机应用》唯一官方网站, 2026, 46(3): 758-766. |
| [8] | 徐千惠, 钮可, 朱顺哲, 石林, 李军. 增强型可逆神经网络视频隐写网络GAB3D-SEVSN[J]. 《计算机应用》唯一官方网站, 2026, 46(2): 467-474. |
| [9] | 吴俊锐, 杨江川, 喻海生, 邹赛, 汪文勇. 基于复增强注意力机制图神经网络的确定性网络性能评估方法[J]. 《计算机应用》唯一官方网站, 2026, 46(2): 505-517. |
| [10] | 罗虎, 张明书. 基于跨模态注意力机制与对比学习的谣言检测方法[J]. 《计算机应用》唯一官方网站, 2026, 46(2): 361-367. |
| [11] | 韦廷浒, 刘皓琰, 吴建宁. 图时空学习模型检测下肢不等长动态变化[J]. 《计算机应用》唯一官方网站, 2026, 46(2): 587-595. |
| [12] | 张日丰, 李广明, 欧阳裕荣. 反射先验图引导的低光图像增强网络[J]. 《计算机应用》唯一官方网站, 2026, 46(2): 546-554. |
| [13] | 林金娇, 张灿舜, 陈淑娅, 王天鑫, 连剑, 徐庸辉. 基于改进图注意力网络的车险欺诈检测方法[J]. 《计算机应用》唯一官方网站, 2026, 46(2): 437-444. |
| [14] | 李名, 王孟齐, 张爱丽, 任花, 窦育强. 基于条件生成对抗网络和混合注意力机制的图像隐写方法[J]. 《计算机应用》唯一官方网站, 2026, 46(2): 475-484. |
| [15] | 张四中, 刘建阳, 李林峰. 基于X3D的轨迹引导感知学习的动作质量评估模型[J]. 《计算机应用》唯一官方网站, 2026, 46(2): 555-563. |
| 阅读次数 | ||||||
|
全文 |
|
|||||
|
摘要 |
|
|||||
