School of Management Sciences & Information Engineering, Hebei University of Economics & Business, Shijiazhuang 050067, China
Show less
文章历史+
Received
Published
Revised
2025-02-19
2028-07-28
2025-10-27
Online Date
2026-09-20
PDF (8720K)
摘要
关键节点识别是复杂网络研究中的重要课题。传统方法存在节点差异性区分度不足、评估维度单一等问题,基于机器学习的方法则面临运行时间长和训练成本较高的挑战。针对这些问题,提出一种“预训练-自适应迁移-预测”的三阶段递进式学习模型GAMAL(graph adaptive multi-stage active learning),实现关键节点识别。该模型首先基于改进的图注意力网络(GATv2),构建多尺度编码器,聚合局部邻域与全局拓扑特征,并利用合成网络预训练模型,建立初始特征映射关系;然后,通过AP(affinity propagation)聚类和最大不确定采样的主动学习策略,选择真实网络中代表节点,实现模型的自适应迁移;最后,根据模型预测真实网络中节点的影响力分数,识别关键节点。在10个真实网络的实验结果表明,该方法在区分度、准确率、运行时间方面表现出色能够准确、高效识别复杂网络的关键节点。
Abstract
Critical node identification is an important topic in the study of complex networks. Traditional methods have problems such as insufficient node differentiation and single evaluation dimension, while machine learning-based methods face the challenges of long running time and high training cost. In order to solve these problems, a three-stage progressive learning model (GAMAL) with “pre-training-adaptive transfer-prediction” was proposed to realize the identification of key nodes. Firstly, based on the improved graph attention network (GATv2), a multi-scale encoder was constructed to aggregate local neighborhood and global topological features, and the synthetic network pre-training model was used to establish the initial feature mapping relationship. Then, through the active learning strategy of AP(affinity propagation) clustering and maximum uncertainty sampling, representative nodes in the real network were selected to realize the adaptive migration of the model. Finally, the influence score of the nodes in the real network was predicted according to the model, and the key nodes were identified. Experimental results on 10 real networks show that the proposed method performs well in terms of discrimination, accuracy and running time, and can accurately and efficiently identify the key nodes of complex networks.
WANG Yue, LIU Yuan-ying, ZHANG Peng-yun, XIAO Xiao-chan.
Identification of Key Nodes in Complex Networks Based on Graph Attention Mechanism and Adaptive Transfer[J].
Science Technology and Engineering, 2026, 26(3): 1098-1106 DOI:10.12404/j.issn.1671-1815.2501038
综上所述,目前的关键节点识别方法在复杂网络中还存在节点表征不足、准确率低、泛化能力差的挑战,难以适应真实网络。基于此,提出一种将图注意力机制与迁移学习和主动学习融合的三阶段递进式模型GAMAL(graph attention mechanism with transfer learning and active learning)。该模型利用改进的图注意力网络,根据节点间相关性动态捕捉节点的局部结构和全局信息,充分考虑节点和邻居之间的交互及网络中的关键信息,为模型训练提供良好的基础。并在真实网络中采用基于AP (affinity propagation)聚类和最大不确定性采样的自适应迁移策略,实现跨网络迁移的精确估计。本文方法在提高模型训练效率的同时,保证模型的泛化能力。使其能在多样化的复杂网络中保持准确性与稳定性。
SouzaR C D, FigueiredoD R, RochaA A D A, et al. Efficient network seeding under variable node cost and limited budget for social networks[J]. Elsevier, 2020, 514: 369-384.
ZhangYongqiang, LiShuang, MaJinlong. Traffic-driven epidemic spread under different routing strategies based on complex networks[J]. Science Technology and Engineering, 2023, 23(4): 1370-1377.
[4]
ZhongS, ZhangH, DengY. Identification of influential nodes in complex networks: a local degree dimension approach[J]. Information Sciences, 2022, 610: 994-1009.
[5]
RuanY R, TangJ, HuY L, et al. Efficient algorithm for the identification of node significance in complex network[J]. IEEE Access, 2020, 8: 28947-28955.
WangFei, WeiLinlin. Fractal characteristics analysis of air traffic flow based on complex network[J]. Science Technology and Engineering, 2023, 23(9): 3982-3990.
ZhangRui, ZhuChunyan, WangQiong, et al. Analysis on the characteristics of airline network in China’s four major multiple-airport systems based on complex network[J]. Science Technology and Engineering, 2023, 23(18): 8002-8010.
[10]
GuetticheM, KheddouciH. Critical links detection in stochastic networks: application to the transport networks[J]. International Journal of Intelligent Computing and Cybernetics, 2019, 12(1): 42-69.
[11]
KitsakM, GallosL, HavlinS, et al. Identification of influential spreaders in complex networks[J]. Nature Physics, 2010, 6(11): 888-893.
[12]
HosniA I E, LiK, AhmadS. Minimizing rumor influence in multiplex online social networks based on human individual and social behaviors[J]. Information Sciences, 2020, 512: 1458-1480.
[13]
ZareieA, SheikhahmadiA. A hierarchical approach for influential node ranking in complex social network[J]. Expert Systems with Aplications, 2018, 93: 200-211.
[14]
TengX, PeiS, MoroneF, et al. Collective influence of multiple spreaders evaluated by tracing real information flow in large-scalesocial networks[J]. Scientific Reports, 2016, 6(1): 36043.
XiongCaiquan, GuXiaohui, WuXinyun. Evaluation method of node importance in complex networks based on K-shell position and two-hop neighbors[J]. Application Research of Computers, 2023, 40(3): 738-742.
CuiXuelian, DongHaitao. Research on social network public opinion control strategy based on key node identification[J]. Computer Engineering and Application, 2023, 10(17): 1-14.
WuBo, LiangXun, ZhangShusen, et al. Advances and applications in graph neural network[J]. Chinese Journal of Computers, 2022, 45(1): 35-68.
[27]
YuE Y, WangY P, FuY, et al. Identifying critical nodes in complex networks via graph convolutional networks[J]. Knowledge-Based Systems, 2020, 198: 105893.
[28]
KumarS, MallikA, PandaB S. Influence maximization in social networks using transfer learning via graph-based LSTM[J]. Expert Systems with Applications, 2023, 212: 118770.
ZhangMinglei, SongYurong, QuHongbo. Identification of key nodes in complex networks based on graph attention mechanism[J]. Complex Systems and Complexity Science, 2024, 17(2): 1-8.
YangYang, WangJunfeng. Research on key node identification in complex networks based on GCN[J]. Journal of Sichuan University (Natural Science Edition), 2023, 60(3): 55-64.
[33]
ZhouS B, ChenZ W, RaoD, et al. Multi-exemplar affinity propagation clustering based on local density peak[J]. Applied Intelligence, 2024, 54(3): 2915-2939.
[34]
AhmadA R, JustinM, MahdiJ, et al. A machine learning-based approach for vital node identification in complex networks[J]. Expert Systems with Applications, 2023, 214: 119086.
[35]
VrahatisA G, LazarosK, KotsiantisS K, et al. Graph attention networks: a comprehensive review of methods and applications[J]. Future Internet, 2024, 16(9): 318.
QianRong, WangJiarui, XingFangyuan, et al. Discovery of important nodes in social networks based on propagation attributes[J]. Computer Engineering and Applications, 2022, 58(16): 184-193.
[38]
GaoS, MaJ, ChenZ, et al. Ranking the spreading ability of nodes in complex networks based on local structure[J]. Physica A: Statistical Mechanics and Its Applications, 2014, 403: 130-147.
SunBaibing, SunJiazheng, HeQuan, et al. Node importance analysis of integration community assessment[J]. Computer Engineering and Applications, 2023, 59(3): 226-233.