Node order is one of the most important factors in learning the structure of a Bayesian network(BN)for probabilistic reasoning.To improve the BN structure learning,we propose a node order learning algorithmbased on th...Node order is one of the most important factors in learning the structure of a Bayesian network(BN)for probabilistic reasoning.To improve the BN structure learning,we propose a node order learning algorithmbased on the frequently used Bayesian information criterion(BIC)score function.The algorithm dramatically reduces the space of node order and makes the results of BN learning more stable and effective.Specifically,we first find the most dependent node for each individual node,prove analytically that the dependencies are undirected,and then construct undirected subgraphs UG.Secondly,the UG-is examined and connected into a single undirected graph UGC.The relation between the subgraph number and the node number is analyzed.Thirdly,we provide the rules of orienting directions for all edges in UGC,which converts it into a directed acyclic graph(DAG).Further,we rank the DAG’s topology order and describe the BIC-based node order learning algorithm.Its complexity analysis shows that the algorithm can be conducted in linear time with respect to the number of samples,and in polynomial time with respect to the number of variables.Finally,experimental results demonstrate significant performance improvement by comparing with other methods.展开更多
Based on a node group <img src="Edit_effba4ca-e855-418a-8a72-d70cb1ec3470.png" width="240" height="46" alt="" />, the Newman type rational operator is constructed in the p...Based on a node group <img src="Edit_effba4ca-e855-418a-8a72-d70cb1ec3470.png" width="240" height="46" alt="" />, the Newman type rational operator is constructed in the paper. The convergence rate of approximation to a class of non-smooth functions is discussed, which is <img src="Edit_174e8f70-651b-4abb-a8f3-a16a576536dc.png" width="85" height="50" alt="" /> regarding to X. Moreover, if the operator is constructed based on further subdivision nodes, the convergence rate is <img src="Edit_557b3a01-7f56-41c0-bb67-deab88b9cc63.png" width="85" height="45" alt="" />. The result in this paper is superior to the approximation results based on equidistant nodes, Chebyshev nodes of the first kind and Chebyshev nodes of the second kind.展开更多
网络切片技术旨在共享的物理网络上创建出多个满足不同业务场景的虚拟网络,基于MCTS(Monte Carlo Tree Search)的切片部署方法在接受率等多种指标上均优于传统的启发式算法。针对基于MCTS的切片部署方法在大规模网络上存在的收敛时间过...网络切片技术旨在共享的物理网络上创建出多个满足不同业务场景的虚拟网络,基于MCTS(Monte Carlo Tree Search)的切片部署方法在接受率等多种指标上均优于传统的启发式算法。针对基于MCTS的切片部署方法在大规模网络上存在的收敛时间过长的问题,论文提出利用通过限制两跳之间的路径长度、并使用节点重要性排序结果作为先验知识对搜索树进行剪枝解决问题。进一步,论文使用广义网络温度GNT(Generalized Network Temperature)对MCTS的搜索目标进行改进。实验结果表明,论文提出的方法有效减缓了因为虚拟网络部署导致的底层网络资源碎片化问题。展开更多
图池化作为图神经网络中重要的组件,在获取图的多粒度信息的过程中扮演了重要角色。而当前的图池化操作均以平等地位看待数据点,普遍未考虑利用邻域内数据之间的偏序关系,从而造成图结构信息破坏。针对此问题,本文提出一种基于偏序关系...图池化作为图神经网络中重要的组件,在获取图的多粒度信息的过程中扮演了重要角色。而当前的图池化操作均以平等地位看待数据点,普遍未考虑利用邻域内数据之间的偏序关系,从而造成图结构信息破坏。针对此问题,本文提出一种基于偏序关系的多视图多粒度图表示学习框架(multi-view and multi-granularity graph representation learning based on partial order relationships,MVMGr-PO),它通过从节点特征视图、图结构视图以及全局视图对节点进行综合评分,进而基于节点之间的偏序关系进行下采样操作。相比于其他图表示学习方法,MVMGr-PO可以有效地提取多粒度图结构信息,从而可以更全面地表征图的内在结构和属性。此外,MVMGr-PO可以集成多种图神经网络架构,包括GCN(graph convolutional network)、GAT(graph attention network)以及GraphSAGE(graph sample and aggregate)等。通过在6个数据集上进行实验评估,与现有基线模型相比,MVMGr-PO在分类准确率上有明显提升。展开更多
基金The work partially supported by the National Natural Science Foundation of China(Grant Nos.61432011,U1435212,61322211 and 61672332)the Postdoctoral Science Foundation of China(2016M591409)+1 种基金the Natural Science Foundation of Shanxi Province,China(201801D121115 and 2013011016-4)Research Project Supported by Shanxi Scholarship Council of China(2020-095).
文摘Node order is one of the most important factors in learning the structure of a Bayesian network(BN)for probabilistic reasoning.To improve the BN structure learning,we propose a node order learning algorithmbased on the frequently used Bayesian information criterion(BIC)score function.The algorithm dramatically reduces the space of node order and makes the results of BN learning more stable and effective.Specifically,we first find the most dependent node for each individual node,prove analytically that the dependencies are undirected,and then construct undirected subgraphs UG.Secondly,the UG-is examined and connected into a single undirected graph UGC.The relation between the subgraph number and the node number is analyzed.Thirdly,we provide the rules of orienting directions for all edges in UGC,which converts it into a directed acyclic graph(DAG).Further,we rank the DAG’s topology order and describe the BIC-based node order learning algorithm.Its complexity analysis shows that the algorithm can be conducted in linear time with respect to the number of samples,and in polynomial time with respect to the number of variables.Finally,experimental results demonstrate significant performance improvement by comparing with other methods.
文摘Based on a node group <img src="Edit_effba4ca-e855-418a-8a72-d70cb1ec3470.png" width="240" height="46" alt="" />, the Newman type rational operator is constructed in the paper. The convergence rate of approximation to a class of non-smooth functions is discussed, which is <img src="Edit_174e8f70-651b-4abb-a8f3-a16a576536dc.png" width="85" height="50" alt="" /> regarding to X. Moreover, if the operator is constructed based on further subdivision nodes, the convergence rate is <img src="Edit_557b3a01-7f56-41c0-bb67-deab88b9cc63.png" width="85" height="45" alt="" />. The result in this paper is superior to the approximation results based on equidistant nodes, Chebyshev nodes of the first kind and Chebyshev nodes of the second kind.
文摘网络切片技术旨在共享的物理网络上创建出多个满足不同业务场景的虚拟网络,基于MCTS(Monte Carlo Tree Search)的切片部署方法在接受率等多种指标上均优于传统的启发式算法。针对基于MCTS的切片部署方法在大规模网络上存在的收敛时间过长的问题,论文提出利用通过限制两跳之间的路径长度、并使用节点重要性排序结果作为先验知识对搜索树进行剪枝解决问题。进一步,论文使用广义网络温度GNT(Generalized Network Temperature)对MCTS的搜索目标进行改进。实验结果表明,论文提出的方法有效减缓了因为虚拟网络部署导致的底层网络资源碎片化问题。
文摘图池化作为图神经网络中重要的组件,在获取图的多粒度信息的过程中扮演了重要角色。而当前的图池化操作均以平等地位看待数据点,普遍未考虑利用邻域内数据之间的偏序关系,从而造成图结构信息破坏。针对此问题,本文提出一种基于偏序关系的多视图多粒度图表示学习框架(multi-view and multi-granularity graph representation learning based on partial order relationships,MVMGr-PO),它通过从节点特征视图、图结构视图以及全局视图对节点进行综合评分,进而基于节点之间的偏序关系进行下采样操作。相比于其他图表示学习方法,MVMGr-PO可以有效地提取多粒度图结构信息,从而可以更全面地表征图的内在结构和属性。此外,MVMGr-PO可以集成多种图神经网络架构,包括GCN(graph convolutional network)、GAT(graph attention network)以及GraphSAGE(graph sample and aggregate)等。通过在6个数据集上进行实验评估,与现有基线模型相比,MVMGr-PO在分类准确率上有明显提升。
