基于改进DDQN算法的复杂网络关键节点识别方法  

Method for identifying key nodes in complex networks based on improved DDQN algorithm

作　　者：江宇楠 刘琳岚[1] 舒坚[2] Jiang Yunan;Liu Linlan;Shu Jian(School of Information Engineering,Nanchang Hangkong University,Nanchang 330063,China;School of Software,Nanchang Hangkong University,Nanchang 330063,China)

机构地区：[1]南昌航空大学信息工程学院,南昌330063 [2]南昌航空大学软件学院,南昌330063

出　　处：《计算机应用研究》2025年第4期1122-1127,共6页Application Research of Computers

基　　金：国家自然科学基金资助项目(62062050);江西省研究生创新专项资金资助项目(YC2023-25)。

摘　　要：为全面提取节点的全局特征,提高复杂网络关键节点识别结果的准确性,提出一种基于改进DDQN(double deep Q-network)算法的复杂网络关键节点识别方法。通过重构DDQN的初始奖励值、引入回退探索和优先访问方法,改进DDQN算法,提取节点全局特征,从而提升全局特征提取的效率和提取结果的准确性。引入聚类系数获取节点的局部特征,通过网络性能均值实验得到全局特征和局部特征的融合参数,对全局特征和局部特征进行融合,得到节点的重要度排序,从而实现关键节点识别。在7个真实网络数据集上的实验结果表明,此方法在基于网络性能均值的评价指标以及SIR模型上均优于对比的基线方法。证明其可以更全面地提取节点全局特征,更准确地识别关键节点。To comprehensively extract global features of nodes and enhance the accuracy of identifying key nodes in complex networks,this paper proposed a method based on an improved DDQN algorithm for key node identification in complex networks.By redefining the initial reward values of DDQN,introducing backtracking exploration and priority access methods,it enhanced the DDQN algorithm to extract the global features of nodes,thereby improving the efficiency of global feature extraction and the accuracy of the extracted results.Introducing clustering coefficient to extract the local features of nodes,deriving fusion parameters for global and local features through mean network performance experiments,and integrating global and local features to rank the importance of nodes,thus achieving key node identification.Experimental results on seven real network datasets demonstrate that this method outperforms the baseline methods in terms of evaluation indicator based on mean network performance and the SIR model.This serves as evidence that this method can comprehensively extract global features of nodes and accurately identify key nodes.

关 键 词：复杂网络 关键节点 DDQN算法 回退探索 优先访问 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]