一种基于BiLSTM 的混合层次化图分类模型  

A Hybrid Hierarchical Graph Classification ModelBased on BiLSTM

作　　者：张红梅[1,2] 郑创 钟晓雄 ZHANG Hong-mei;ZHENG Chuang;ZHONG Xiao-xiong(School of Information and Communication,Guilin University of Electronic Technology,Guilin Guangxi 541004,China;Pengcheng Lab,Shenzhen Guangdong 518005,China)

机构地区：[1]桂林电子科技大学信息与通信学院,广西桂林541004 [2]鹏程实验室,广东深圳518055

出　　处：《计算机仿真》2024年第4期260-264,340,共6页Computer Simulation

基　　金：广西区自然科学基金重点项目(2020GXNSFDA238001);鹏程实验室重大攻关项目(PCL2021A02);桂林电子科技大学研究生教育创新计划资助项目(2021YCXS046)。

摘　　要：图分类在化学和生物信息学等诸多领域中是一个非常重要且极具挑战的问题,GNN模型是图分类问题的主流方法。现有的GNN模型采用卷积操作来实现邻域节点信息聚集,再通过池化操作生成粗化图。然而,仅通过池化方法不能捕获到每次卷积后读出图的双向依赖关系。为了提取到更充分的特征信息,提出一种混合层次化模型,首先分别提取节点特征信息和结构特征信息,再将特征信息融合,然后采用BiLSTM捕获不同层次读出图之间的双向依赖关系,从而提取到更丰富的特征信息。实验结果表明,与对比模型相比,上述模型的准确度有着明显的提升。Graph classification is a very important and challenging problem in many fields such as chemistry and bioinformatics,and the GNN model is the mainstream method for graph classification problems.The existing GNN models use convolution operations to aggregate the information of neighbor nodes,and then generate the coarseninggraphs by pooling method.However,the bidirectional dependencies of the readout graph after each convolution cannot be captured by the pooling method alone.In order to extract more sufficient feature information,a hybrid hierarchical model is proposed in this paper,which first extracts node feature information and structural feature information respectively,then fuses the feature information,and then uses BiLSTM to capture the bidirectional dependencies between different levels of readout graphs,so as to extract richer feature information.The experimental results show that compared with the comparison model,the accuracy of the model has been significantly improved。

关 键 词：图神经网络 层次顺序 双向依赖关系 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]