Traffic prediction is a necessary function in intelligent transporta-tion systems to alleviate traffic congestion.Graph learning methods mainly focus on the spatiotemporal dimension,but ignore the nonlinear movement o...Traffic prediction is a necessary function in intelligent transporta-tion systems to alleviate traffic congestion.Graph learning methods mainly focus on the spatiotemporal dimension,but ignore the nonlinear movement of traffic prediction and the high-order relationships among various kinds of road segments.There exist two issues:1)deep integration of the spatiotempo-ral information and 2)global spatial dependencies for structural properties.To address these issues,we propose a nonlinear spatiotemporal optimization method,which introduces hypergraph convolution networks(HGCN).The method utilizes the higher-order spatial features of the road network captured by HGCN,and dynamically integrates them with the historical data to weigh the influence of spatiotemporal dependencies.On this basis,an extended Kalman filter is used to improve the accuracy of traffic prediction.In this study,a set of experiments were conducted on the real-world dataset in Chengdu,China.The result showed that the proposed method is feasible and accurate by two different time steps.Especially at the 15-minute time step,compared with the second-best method,the proposed method achieved 3.0%,11.7%,and 9.0%improvements in RMSE,MAE,and MAPE,respectively.展开更多
现有的电子健康记录(electronic health records,EHR)的图表示学习方法多依赖单个患者的局部信息,忽视了群体患者在疾病演化和诊疗路径上的潜在关联,从而限制了模型的泛化性与鲁棒性.针对这一问题,本文提出一种混合多层级图神经网络(hyb...现有的电子健康记录(electronic health records,EHR)的图表示学习方法多依赖单个患者的局部信息,忽视了群体患者在疾病演化和诊疗路径上的潜在关联,从而限制了模型的泛化性与鲁棒性.针对这一问题,本文提出一种混合多层级图神经网络(hybrid multi-level graph neural network,H-MGNN)模型,并将其应用于重症监护室(intensive care unit,ICU)患者的死亡预测.该模型通过构建宏观层面的患者关系图(patient-patient graph,P-P)、微观层面的分类-笔记-词汇超图(taxonomy-note-word hypergraph,T-N-W),结合超图的时序依赖关系,实现多尺度上的患者特征融合.同时,本文设计了融合算法(hybrid embedding,Hybrid-E),用于提取和整合患者嵌入的潜在特征,以提升预测准确性.实验结果表明,H-MGNN在MIMIC-Ⅲ(medical information mart for intensive care Ⅲ)数据集上的住院死亡率预测等任务中显著优于现有方法,验证了其在复杂EHR数据挖掘中的有效性和先进性.展开更多
The unique ways of information organization and dissemination was examined through the microblog and the real-name social network as the representatives of the new virtual social networks. In order to discuss the inte...The unique ways of information organization and dissemination was examined through the microblog and the real-name social network as the representatives of the new virtual social networks. In order to discuss the interrelation and interaction of the two dimensions-topic and user, a supernetwork model was established based on the supernetwork research method. Through the actual data, a supernetwork topology diagram and the changing rule of user participation were attained. And it was concluded that the key factor of dealing with emergent online public sentiment should start with affecting the opinions of key figures, whose opinions would further affect the public opinions.展开更多
semantics information while maintaining spatial detail con-texts.Long-range context information plays a crucial role in this scenario.How-ever,the traditional convolution kernel only provides the local and small size ...semantics information while maintaining spatial detail con-texts.Long-range context information plays a crucial role in this scenario.How-ever,the traditional convolution kernel only provides the local and small size of the receptivefield.To address the problem,we propose a plug-and-play module aggregating both local and global information(aka LGIA module)to capture the high-order relationship between nodes that are far apart.We incorporate both local and global correlations into hypergraph which is able to capture high-order rela-tionships between nodes via the concept of a hyperedge connecting a subset of nodes.The local correlation considers neighborhood nodes that are spatially adja-cent and similar in the same CNN feature maps of magnetic resonance(MR)image;and the global correlation is searched from a batch of CNN feature maps of MR images in feature space.The influence of these two correlations on seman-tic segmentation is complementary.We validated our LGIA module on various CNN segmentation models with the cardiac MR images dataset.Experimental results demonstrate that our approach outperformed several baseline models.展开更多
The concepts of modified graphs of a composite graph with respect to two vertex-pairs and a hyperedge-decomposition are introduced,respectively.By applying them and thedirected hypergraph theory,the topological formul...The concepts of modified graphs of a composite graph with respect to two vertex-pairs and a hyperedge-decomposition are introduced,respectively.By applying them and thedirected hypergraph theory,the topological formulas for the parameter-extraction theorem andsubnetwork-extraction theorems are derived,and then the topological formulas for multiterminalfeedback networks are presented.In these formulas the parameters of the feedback subnetworkare separated from that of the fundamental subnetwork,so that it is convenient to find out theeffect of the feedback parameters.Furthermore,since one network is decomposed into two smallersubnetworks,the computing time complexity and space complexity can be reduced.展开更多
First,the state space tree method for finding communication network overall re-liability is presented.It directly generates one disjoint tree multilevel polynomial of a networkgraph.Its advantages are smaller computat...First,the state space tree method for finding communication network overall re-liability is presented.It directly generates one disjoint tree multilevel polynomial of a networkgraph.Its advantages are smaller computational effort(its computing time complexity is O(en_l),where e is the number of edges and n_l is the number of leaves)and shorter resulting expression.Second,based on it an exact decomposition algorithm for finding communication network overallreliability is presented by applying the hypergraph theory.If we use it to carry out the m-timedecomposition of a network graph,the communication network scale which can be analyzed by acomputer can be extended to m-fold.展开更多
文摘Traffic prediction is a necessary function in intelligent transporta-tion systems to alleviate traffic congestion.Graph learning methods mainly focus on the spatiotemporal dimension,but ignore the nonlinear movement of traffic prediction and the high-order relationships among various kinds of road segments.There exist two issues:1)deep integration of the spatiotempo-ral information and 2)global spatial dependencies for structural properties.To address these issues,we propose a nonlinear spatiotemporal optimization method,which introduces hypergraph convolution networks(HGCN).The method utilizes the higher-order spatial features of the road network captured by HGCN,and dynamically integrates them with the historical data to weigh the influence of spatiotemporal dependencies.On this basis,an extended Kalman filter is used to improve the accuracy of traffic prediction.In this study,a set of experiments were conducted on the real-world dataset in Chengdu,China.The result showed that the proposed method is feasible and accurate by two different time steps.Especially at the 15-minute time step,compared with the second-best method,the proposed method achieved 3.0%,11.7%,and 9.0%improvements in RMSE,MAE,and MAPE,respectively.
文摘现有的电子健康记录(electronic health records,EHR)的图表示学习方法多依赖单个患者的局部信息,忽视了群体患者在疾病演化和诊疗路径上的潜在关联,从而限制了模型的泛化性与鲁棒性.针对这一问题,本文提出一种混合多层级图神经网络(hybrid multi-level graph neural network,H-MGNN)模型,并将其应用于重症监护室(intensive care unit,ICU)患者的死亡预测.该模型通过构建宏观层面的患者关系图(patient-patient graph,P-P)、微观层面的分类-笔记-词汇超图(taxonomy-note-word hypergraph,T-N-W),结合超图的时序依赖关系,实现多尺度上的患者特征融合.同时,本文设计了融合算法(hybrid embedding,Hybrid-E),用于提取和整合患者嵌入的潜在特征,以提升预测准确性.实验结果表明,H-MGNN在MIMIC-Ⅲ(medical information mart for intensive care Ⅲ)数据集上的住院死亡率预测等任务中显著优于现有方法,验证了其在复杂EHR数据挖掘中的有效性和先进性.
基金National Natural Science Foundation of China (No. 71071098)
文摘The unique ways of information organization and dissemination was examined through the microblog and the real-name social network as the representatives of the new virtual social networks. In order to discuss the interrelation and interaction of the two dimensions-topic and user, a supernetwork model was established based on the supernetwork research method. Through the actual data, a supernetwork topology diagram and the changing rule of user participation were attained. And it was concluded that the key factor of dealing with emergent online public sentiment should start with affecting the opinions of key figures, whose opinions would further affect the public opinions.
基金supported by the Sichuan Science and Technology Program(Grant No.2019ZDZX0005,2019YFG0496,2020YFG0143,2019JDJQ0002 and 2020YFG0009).
文摘semantics information while maintaining spatial detail con-texts.Long-range context information plays a crucial role in this scenario.How-ever,the traditional convolution kernel only provides the local and small size of the receptivefield.To address the problem,we propose a plug-and-play module aggregating both local and global information(aka LGIA module)to capture the high-order relationship between nodes that are far apart.We incorporate both local and global correlations into hypergraph which is able to capture high-order rela-tionships between nodes via the concept of a hyperedge connecting a subset of nodes.The local correlation considers neighborhood nodes that are spatially adja-cent and similar in the same CNN feature maps of magnetic resonance(MR)image;and the global correlation is searched from a batch of CNN feature maps of MR images in feature space.The influence of these two correlations on seman-tic segmentation is complementary.We validated our LGIA module on various CNN segmentation models with the cardiac MR images dataset.Experimental results demonstrate that our approach outperformed several baseline models.
文摘多行为推荐(multi-behavior recommendation,MBR)在互联网平台中愈发重要,但现有方法仍面临两大挑战:a)无法刻画用户不同行为下的复杂兴趣偏好;b)难以建模不同行为间的相互关系。基于此,提出一种对比学习增强的多行为超图神经网络模型(multi-behavior hypergraph neural network model enhanced with contrastive lear-ning,MBHCL),在建模用户复杂多类型交互的同时,结合对比学习捕获行为间共性与差异,以获取更优嵌入表示,缓解冷启动与数据稀疏问题。具体地,MBHCL首先构建用户-项目多行为交互超图,以刻画用户对项目不同维度的偏好;其次设计三个对比任务整合单行为表示,通过捕捉行为间的共性与差异获取全面用户兴趣偏好。最终,MBHCL在四个真实场景数据集上进行对比实验。结果表明,在Tmall和BeiBei数据集上,HIT和NDCG指标有至少4.8%的提升,在Kuairand和Yelp数据集上,HIT和NDCG指标至少提升3.6%,并通过消融实验验证了各模块的有效性,同时显著改善了冷启动用户推荐效果。
文摘The concepts of modified graphs of a composite graph with respect to two vertex-pairs and a hyperedge-decomposition are introduced,respectively.By applying them and thedirected hypergraph theory,the topological formulas for the parameter-extraction theorem andsubnetwork-extraction theorems are derived,and then the topological formulas for multiterminalfeedback networks are presented.In these formulas the parameters of the feedback subnetworkare separated from that of the fundamental subnetwork,so that it is convenient to find out theeffect of the feedback parameters.Furthermore,since one network is decomposed into two smallersubnetworks,the computing time complexity and space complexity can be reduced.
文摘First,the state space tree method for finding communication network overall re-liability is presented.It directly generates one disjoint tree multilevel polynomial of a networkgraph.Its advantages are smaller computational effort(its computing time complexity is O(en_l),where e is the number of edges and n_l is the number of leaves)and shorter resulting expression.Second,based on it an exact decomposition algorithm for finding communication network overallreliability is presented by applying the hypergraph theory.If we use it to carry out the m-timedecomposition of a network graph,the communication network scale which can be analyzed by acomputer can be extended to m-fold.
