In this paper,large deviations principle(LDP)and moderate deviations principle(MDP)of record numbers in random walks are studied under certain conditions.The results show that the rate functions of LDP and MDP are dif...In this paper,large deviations principle(LDP)and moderate deviations principle(MDP)of record numbers in random walks are studied under certain conditions.The results show that the rate functions of LDP and MDP are different from those of weak record numbers,which are interesting complements of the conclusions by Li and Yao[1].展开更多
Missiles provide long-range precision strike capabilities and have become a cornerstone of modern warfare.The contrail clouds formed by missile during their active flight phase present significant chal-lenges to high-...Missiles provide long-range precision strike capabilities and have become a cornerstone of modern warfare.The contrail clouds formed by missile during their active flight phase present significant chal-lenges to high-altitude environmental observation and target detection and tracking.Existing studies primarily focus on specific airspace regions,leaving critical gaps in understanding the effects of long dispersion times,wide altitude ranges,and variable atmospheric conditions on missile contrail clouds.To address these gaps,this article develops a numerical method based on the Lagrangian random walk model,which incorporates various velocity variation terms,including particle velocity caused by the difference of wind field,by the thermal motion of local gas molecules and by random collisions between contrail cloud particles to capture the influence of environmental wind fields,atmospheric conditions,and particle concentrations on the motion of contrail cloud particles.A general coordinate system aligned with the missile's flight trajectory is employed to represent particle distribution characteristics.The proposed method is in good agreement with the conducted experiments as well as with the available numerical simulations.The results demonstrate that the proposed model effectively simulates the dispersion state of contrail clouds,accurately reflecting the impact of large-scale wind field variations and altitude changes with high computational efficiency.Additionally,simulation results indicate that the increased distance between gas molecules in rarefied environments facilitates enhanced particle dispersion,while larger particles exhibit a faster dispersion rate due to their greater mass.展开更多
Unstructured P2P has power-law link distribution, and the random walk in power-law networks is analyzed. The analysis results show that the probability that a random walker walks through the high degree nodes is high ...Unstructured P2P has power-law link distribution, and the random walk in power-law networks is analyzed. The analysis results show that the probability that a random walker walks through the high degree nodes is high in the power-law network, and the information on the high degree nodes can be easily found through random walk. Random walk spread and random walk search method （RWSS） is proposed based on the analysis result. Simulation results show that RWSS achieves high success rates at low cost and is robust to high degree node failure.展开更多
Researchers face many class prediction challenges stemming from a small size of training data vis-a-vis a large number of unlabeled samples to be predicted. Transductive learning is proposed to utilize information abo...Researchers face many class prediction challenges stemming from a small size of training data vis-a-vis a large number of unlabeled samples to be predicted. Transductive learning is proposed to utilize information about unlabeled data to estimate labels of the unlabeled data for this condition. This work presents a new transductive learning method called two-way Markov random walk(TMRW) algorithm. The algorithm uses information about labeled and unlabeled data to predict the labels of the unlabeled data by taking random walks between the labeled and unlabeled data where data points are viewed as nodes of a graph. The labeled points correlate to unlabeled points and vice versa according to a transition probability matrix. We can get the predicted labels of unlabeled samples by combining the results of the two-way walks. Finally, ensemble learning is combined with transductive learning, and Adboost.MH is taken as the study framework to improve the performance of TMRW, which is the basic learner. Experiments show that this algorithm can predict labels of unlabeled data well.展开更多
社交网络中,节点间存在多种关系类型,节点数量会随着时间的推移而变化,这种异质性和动态性给链路预测任务带来极大的挑战。因此,本文提出一种基于增量学习的社交网络链路预测方法(incremental learning social networks link prediction...社交网络中,节点间存在多种关系类型,节点数量会随着时间的推移而变化,这种异质性和动态性给链路预测任务带来极大的挑战。因此,本文提出一种基于增量学习的社交网络链路预测方法(incremental learning social networks link prediction,IL-SNLP)。通过对网络进行分层,使每一层网络只包含一种关系类型,以更好地获取节点在每种关系类型下的语义信息;针对网络的动态性,利用时序随机游走捕获社交网络中的局部结构信息和时序信息;针对增量数据,采用增量式更新随机游走策略对历史随机游走序列进行更新。通过增量式skip-gram模型从随机游走序列中提取新出现节点的特征,并进一步更新历史节点的特征;针对网络的异质性,采用概率模型提取不同关系类型之间的因果关系关联程度,并将其作用于每一层的节点特征,以改善不同关系层下节点特征表现能力;利用多层感知机构建节点相互感知器,挖掘节点间建立连接时的相互贡献,实现更高的链路预测准确率。实验结果表明,在3个真实的社交网络数据集上,IL-SNLP方法的ROC曲线下的面积(AUC)和F1分数比基线方法分别提高了10.08%~67.60%和1.76%~64.67%,提升了预测性能;对于增量数据,只需要少次迭代就能保持预测模型的性能,提高了模型训练的速度;与未采用增量学习技术的IL-SNLP−方法相比,IL-SNLP方法在时间效率上提升了30.78%~257.58%,显著缩短了模型的运行时长。展开更多
局部多重社区发现是社交网络分析中的关键技术,旨在揭示网络中用户的多重归属和复杂联系。针对现有局部多重社区发现算法大多基于网络拓扑结构,忽视节点属性信息的问题,提出了融合节点属性的局部多重社区发现算法(MLCDINA)。该算法将属...局部多重社区发现是社交网络分析中的关键技术,旨在揭示网络中用户的多重归属和复杂联系。针对现有局部多重社区发现算法大多基于网络拓扑结构,忽视节点属性信息的问题,提出了融合节点属性的局部多重社区发现算法(MLCDINA)。该算法将属性网络的结构和属性信息相结合为节点对之间的边权重,并通过随机游走评估节点间结构和属性的融合重要性(IISA)。此外,该算法引入了考虑边权重的局部聚类系数和亲密度随机游走(IRW),以增强对子图稠密性和IISA的评估。实验结果表明,MLCDINA在真实属性网络上的Jaccard F 1-score较现有算法有显著提升,验证了其在局部多重社区发现任务中的有效性。展开更多
基金supported by the National Natural Science Foundation of China(Grant No.11671145)the Science and Technology Commission of Shanghai Municipality(Grant No.18dz2271000).
文摘In this paper,large deviations principle(LDP)and moderate deviations principle(MDP)of record numbers in random walks are studied under certain conditions.The results show that the rate functions of LDP and MDP are different from those of weak record numbers,which are interesting complements of the conclusions by Li and Yao[1].
文摘Missiles provide long-range precision strike capabilities and have become a cornerstone of modern warfare.The contrail clouds formed by missile during their active flight phase present significant chal-lenges to high-altitude environmental observation and target detection and tracking.Existing studies primarily focus on specific airspace regions,leaving critical gaps in understanding the effects of long dispersion times,wide altitude ranges,and variable atmospheric conditions on missile contrail clouds.To address these gaps,this article develops a numerical method based on the Lagrangian random walk model,which incorporates various velocity variation terms,including particle velocity caused by the difference of wind field,by the thermal motion of local gas molecules and by random collisions between contrail cloud particles to capture the influence of environmental wind fields,atmospheric conditions,and particle concentrations on the motion of contrail cloud particles.A general coordinate system aligned with the missile's flight trajectory is employed to represent particle distribution characteristics.The proposed method is in good agreement with the conducted experiments as well as with the available numerical simulations.The results demonstrate that the proposed model effectively simulates the dispersion state of contrail clouds,accurately reflecting the impact of large-scale wind field variations and altitude changes with high computational efficiency.Additionally,simulation results indicate that the increased distance between gas molecules in rarefied environments facilitates enhanced particle dispersion,while larger particles exhibit a faster dispersion rate due to their greater mass.
文摘Unstructured P2P has power-law link distribution, and the random walk in power-law networks is analyzed. The analysis results show that the probability that a random walker walks through the high degree nodes is high in the power-law network, and the information on the high degree nodes can be easily found through random walk. Random walk spread and random walk search method （RWSS） is proposed based on the analysis result. Simulation results show that RWSS achieves high success rates at low cost and is robust to high degree node failure.
基金Project(61232001) supported by National Natural Science Foundation of ChinaProject supported by the Construct Program of the Key Discipline in Hunan Province,China
文摘Researchers face many class prediction challenges stemming from a small size of training data vis-a-vis a large number of unlabeled samples to be predicted. Transductive learning is proposed to utilize information about unlabeled data to estimate labels of the unlabeled data for this condition. This work presents a new transductive learning method called two-way Markov random walk(TMRW) algorithm. The algorithm uses information about labeled and unlabeled data to predict the labels of the unlabeled data by taking random walks between the labeled and unlabeled data where data points are viewed as nodes of a graph. The labeled points correlate to unlabeled points and vice versa according to a transition probability matrix. We can get the predicted labels of unlabeled samples by combining the results of the two-way walks. Finally, ensemble learning is combined with transductive learning, and Adboost.MH is taken as the study framework to improve the performance of TMRW, which is the basic learner. Experiments show that this algorithm can predict labels of unlabeled data well.
文摘社交网络中,节点间存在多种关系类型,节点数量会随着时间的推移而变化,这种异质性和动态性给链路预测任务带来极大的挑战。因此,本文提出一种基于增量学习的社交网络链路预测方法(incremental learning social networks link prediction,IL-SNLP)。通过对网络进行分层,使每一层网络只包含一种关系类型,以更好地获取节点在每种关系类型下的语义信息;针对网络的动态性,利用时序随机游走捕获社交网络中的局部结构信息和时序信息;针对增量数据,采用增量式更新随机游走策略对历史随机游走序列进行更新。通过增量式skip-gram模型从随机游走序列中提取新出现节点的特征,并进一步更新历史节点的特征;针对网络的异质性,采用概率模型提取不同关系类型之间的因果关系关联程度,并将其作用于每一层的节点特征,以改善不同关系层下节点特征表现能力;利用多层感知机构建节点相互感知器,挖掘节点间建立连接时的相互贡献,实现更高的链路预测准确率。实验结果表明,在3个真实的社交网络数据集上,IL-SNLP方法的ROC曲线下的面积(AUC)和F1分数比基线方法分别提高了10.08%~67.60%和1.76%~64.67%,提升了预测性能;对于增量数据,只需要少次迭代就能保持预测模型的性能,提高了模型训练的速度;与未采用增量学习技术的IL-SNLP−方法相比,IL-SNLP方法在时间效率上提升了30.78%~257.58%,显著缩短了模型的运行时长。
文摘局部多重社区发现是社交网络分析中的关键技术,旨在揭示网络中用户的多重归属和复杂联系。针对现有局部多重社区发现算法大多基于网络拓扑结构,忽视节点属性信息的问题,提出了融合节点属性的局部多重社区发现算法(MLCDINA)。该算法将属性网络的结构和属性信息相结合为节点对之间的边权重,并通过随机游走评估节点间结构和属性的融合重要性(IISA)。此外,该算法引入了考虑边权重的局部聚类系数和亲密度随机游走(IRW),以增强对子图稠密性和IISA的评估。实验结果表明,MLCDINA在真实属性网络上的Jaccard F 1-score较现有算法有显著提升,验证了其在局部多重社区发现任务中的有效性。
文摘【目的】知识表示学习(Knowledge Representation Learning,KRL)在跨语言实体对齐方面取得了显著成就,但未能建模异构知识图谱之间的复杂语义关系,且现有方法多数依赖局部特征匹配以至于未能充分利用知识图谱结构信息。提出了一种基于判别式多特征融合的实体对齐框架(Entity Alignment algorithm based on Discriminant Multi-feature Fusion,EA-DMF)。【方法】利用知识图谱中的语义信息、结构信息以及属性信息进行多特征融合,充分挖掘出图谱中的潜在语义信息。具体而言,EA-DMF引入Gromov-Wasserstein距离度量图谱之间的相似性,建立了随机关系游走算法,利用知识图谱中的长期依赖关系丰富了实体的语义信息,并通过高置信度锚节点的迭代更新,将高置信度的局部对齐信息逐步扩展至全局,最终应用多视角最优传输理论融合多个信息特征进而得到对齐实体对集合。【结果】经过在五个实体对齐数据集上的广泛实验,在没有任何监督或超参数调整的情况下,EA-DMF超越多个竞争基线,证明该方法能够更有效准确地进行知识图谱中未知实体的对齐。
