graph

摘要： Let G = (V, E) be a graph and Cm be the cycle graph with m vertices. In this paper, we continued Yeh’s work [1] on the distance labeling of the cycle graph Cm. An n-set distance labeling of a graph G is the labeling of the vertices (with n labels per vertex) of G under certain constraints depending on the distance between each pair of the vertices in G. Following Yeh’s notation [1], the smallest value for the largest label in an n-set distance labeling of G is denoted by λ1(n)(G). Basic results were presented in [1] for λ1(2)(Cm) for all m and λ1(n)(Cm) for some m where n ≥ 3. However, there were still gaps left unstudied due to case-by-case complexities. For these uncovered cases, we proved a lower bound for λ1(n)(Cm). Then we proposed an algorithm for finding an n-set distance labeling for n ≥ 3 based on our proof of the lower bound. We verified every single case for n = 3 up to n = 500 by this same algorithm, which indicated that the upper bound is the same as the lower bound for n ≤ 500.

摘要： Recently,whether the channel prediction can be achieved in diverse communication scenarios by directly utilizing the environment information gained lots of attention due to the environment impacting the propagation characteristics of the wireless channel.This paper presents an environment information-based channel prediction(EICP)method for connecting the environment with the channel assisted by the graph neural networks(GNN).Firstly,the effective scatterers(ESs)producing paths and the primary scatterers(PSs)generating single propagation paths are detected by building the scatterercentered communication environment graphs(SCCEGs),which can simultaneously preserve the structure information and highlight the pending scatterer.The GNN-based classification model is implemented to distinguish ESs and PSs from other scatterers.Secondly,large-scale parameters(LSP)and small-scale parameters(SSP)are predicted by employing the GNNs with multi-target architecture and the graphs of detected ESs and PSs.Simulation results show that the average normalized mean squared error(NMSE)of LSP and SSP predictions are 0.12 and 0.008,which outperforms the methods of linear data learning.

摘要： A distributed information network with complex network structure always has a challenge of locating fault root causes.In this paper,we propose a novel root cause analysis(RCA)method by random walk on the weighted fault propagation graph.Different from other RCA methods,it mines effective features information related to root causes from offline alarms.Combined with the information,online alarms and graph relationship of network structure are used to construct a weighted graph.Thus,this approach does not require operational experience and can be widely applied in different distributed networks.The proposed method can be used in multiple fault location cases.The experiment results show the proposed approach achieves much better performance with 6%higher precision at least for root fault location,compared with three baseline methods.Besides,we explain how the optimal parameter’s value in the random walk algorithm influences RCA results.

摘要： SpacetimeAI and GeoAI are currently hot topics,applying the latest algorithms in computer science,such as deep learning,to spatiotemporal data.Although deep learning algorithms have been successfully applied to raster data due to their natural applicability to image processing,their applications in other spatial and space-time data types are still immature.This paper sets up the proposition of using a network(&graph)-based framework as a generic spatial structure to present space-time processes that are usually represented by the points,polylines,and polygons.We illustrate network and graph-based SpaceTimeAI,from graph-based deep learning for prediction,to space-time clustering and optimisation.These applications demonstrate the advantages of network(graph)-based SpacetimeAI in the fields of transport&mobility,crime&policing,and public health.

摘要： TigerGraph创始人兼CEO许昱认为:“图数据库和分析将是商业智能领域数据应用的必然趋势,代表着人工智能时代实现增强分析和机器学习的技术创新和突破。”

摘要： Molecular dynamics is a powerful simulation tool to explore material properties.Most realistic material systems are too large to be simulated using first-principles molecular dynamics.Classical molecular dynamics has a lower computational cost but requires accurate force fields to achieve chemical accuracy.In this work,we develop a symmetry-adapted graph neural network framework called the molecular dynamics graph neural network(MDGNN)to construct force fields automatically for molecular dynamics simulations for both molecules and crystals.This architecture consistently preserves translation,rotation,and permutation invariance in the simulations.We also propose a new feature engineering method that includes high-order terms of interatomic distances and demonstrate that the MDGNN accurately reproduces the results of both classical and first-principles molecular dynamics.In addition,we demonstrate that force fields constructed by the proposed model have good transferability.The MDGNN is thus an efficient and promising option for performing molecular dynamics simulations of large-scale systems with high accuracy.

摘要： For two graphs G and H, if G and H have the same matching polynomial, then G and H are said to be matching equivalent. We denote by δ (G), the number of the matching equivalent graphs of G. In this paper, we give δ (sK1 ∪ t1C9 ∪ t2C15), which is a generation of the results of in [1].

摘要： 随着智能电网建设的不断深入,在配用电环节收集的监测数据越来越多,逐渐构成智能电网用户侧大数据.传统数据分析模式已经无法满足性能需求,迫切需要新的存储和数据分析模式来应对.提出基于阿里云大数据分析平台MaxCompute的海量用电数据聚类分析方法,该方法充分考虑用电数据的特点,设计基于多级分区表的用电数据存储模式,采用三相电压、三相电流、三相功率因数等建立多维数据特征,应用MaxCompute Graph框架设计实现高效的海量用电数据的聚类划分算法.实验结果表明,所设计的存储模式可有效提升用电数据的检索效率;通过对不同用电类型的用户进行聚类划分,聚类准确率达到88％,验证了聚类划分的有效性和高性能.

摘要： Graphcore发布第二代IPU及IPU-M20002020年7月15日,Graphcore正式发布第二代IPU以及用于大规模系统级产品的IPU-Machine:M2000(IPU-M2000),新一代产品具有更强的处理能力、更多的内存和内置的可扩展性,可处理极其庞大的机器智能工作负载。IPU-M2000是一款即插即用的机器智能刀片式计算单元,由Graphcore全新的7纳米Colossus第二代GC200IPU提供动力,并由Poplar软件栈提供全面支持。其设计便于部署,并支持可扩展至大规模的系统。这款纤薄的1U刀片机可提供1个PetaFlop的机器智能计算,并集成了针对AI扩展优化的网络技术。

摘要： Graphcore是一家总部位于英国的创新公司,其主要业务是研发专门应用于AI技术的创新芯片——IPU(Intelligence Processing Unit)。自2016年成立以来,就受到了业界、市场和资本的高度关注。截至目前,Graphcore的总融资额超过4.5亿美金,其全球办公室遍布欧洲、亚洲和北美。随着Graphcore IPU(智能处理器)硬件及其开发软件Poplar在人工智能行业的日益升温,日前,Graphcore又发布了Graphcore IPU的第二代产品Colossus Mk2 GC200,以及可以用于大规模系统级产品的IPUMachine:M2000(IPU-M2000)。第二代IPU具有更强的处理能力、更多的内存和内置的可扩展性,可处理庞大的机器智能工作负载。

摘要： As XML data becomes more and more, graph structure join urgently requires rapid judgement algorithm of the reachability of the two vertexes in a graph. The existing graph reachability algorithm includes 2-HOP, HOPI and Dual Labeling, but they don't support graph update, and the time complexity and the space complexity of the first two algorithm are too large to use in the real applications. In the present paper, a new update-aware graph reachability algorithm GraphReach is proposed. The algorithm first computes the minimal equivalent graph of the original graph, sperates it into the spanning tree and non-tree edges. Then the algorithm codes the nodes of the spanning tree and builds index for the non-tree edges. Finally, the algorithm proposes rules to judge the reaachability of the two points in the original graph. Experimental results show that the algorithm proposed in this paper can processing graph reachability fast and the algorithm is useful for real applications.

摘要： Testing reachability between nodes in a graph is a well-known problem with many important applications, including knowledge representation, program analysis, and more recently, ontology management as well as XML query processing. Various approaches have been proposed to encode graph reachability information using node labeling schemes, but they are all unable to effectively support the dynamic update of graph. In this paper, we propose a novel approach called FLS (Fractional Labeling Scheme) to solve this problem, and corresponding algorithms. We also prove some theorems that present analytical results. A notable feature of FLS is that it uses the idea of trichotomy and the unlimited partition nature of fraction number to support fully update of the graph.

摘要： 本发明提供了一种基于深度强化学习的Graph2Seq自动生成Java代码注释的方法。在编码器端，通过将代码首先转化为代码图的形式，利用图结构来表达出代码的结构化信息，再使用图神经网络(Graph Neural Network)对程序图进行编码得到代码的结构化语义信息。在解码器端，本发明使用基于深度强化学习框架改进的解码器得到其对应的自然语言注释内容，解决了传统方法中在推理阶段没有真实序列单词指导导致的高偏差问题，从而实现代码注释自然生成的准确和自然。

摘要： 本发明公开了一种基于融合多注意力机制和Graph的视频语义分析方法。该方法包括首先从视频中提取图像数据；在残差网络的基础上融合基于卷积注意力模块的注意力机制，以提取图像特征；利用文本中的知识预训练出一个具有先验知识的字典；通过基于多图卷机网络将提取的图像特征和中的目标关系属性信息转化成图调制特征；再将图调制特征导入长短时记忆网络生成描述的句子；最终得到视频语义文本内容数据，输出该数据。本发明最终能够得到准确度较高的视频语义文本数据。本发明结合硬件实体的监控设备，可以为公共安全提供更多智能视频分析和提前预警功能，极大程度上提高视频监控的效率，发挥视频资源的最大用途。

摘要： 本发明公开的基于Graph Cuts理论的双目立体图像对目标分割方法，首先，输入双目立体图像中的左右两幅图像，通过立体匹配算法计算得到两幅图像的视差图；其次，计算图像的权重信息；然后，根据权重信息构造Graph Cuts方法的图模型，得到立体图像对的分割结果；最后，将分割结果使用膨胀、腐蚀等形态学操作填补分割结果中小的空洞，去除分割结果中小的连通区域，再使用中值滤波进行平滑，得到最终的分割结果。本发明公开的方法相比传统图像分割方法加强了传统方法中构建的邻域系统，提高图像边缘处的分割精度，同时更好地使用了双目图像的深度信息，得到更好的分割结果。

摘要： 本申请涉及一种Nebula Graph图数据库查询方法、系统、电子装置及介质，其中，该Nebula Graph图数据库查询方法包括：根据用户在可视化交互软件上对nGQL语法元素的移动、拖拽、连线和属性设置结果，生成nGQL图查询语句，所述nGQL图查询语句用于供用户复制至图数据库的控制台，以对所述图数据库执行查询；通过本申请，解决了相关技术中编程工程师查询Nebula Graph图数据库的技术门槛较高的问题，在满足了编程人员对Nebula Graph图数据库的查询功能的需求的基础上，降低了编程人员查询Nebula Graph图数据库的技术门槛。

摘要： 本发明公开了基于Graph Cuts算法的漂浮物识别方法，属于图片处理技术领域，该识别方法具体步骤如下：(1)获取漂浮物实景图；(2)图像预处理；(3)图像分割；(4)区域特征提取；(5)生成智能识别模型；(6)漂浮物识别；(7)漂浮物定位；本发明采用Graph Cuts算法作为漂浮物实景图像分割工具，其具有精度高、鲁棒性强和运行效率快的特点，进而有利于为后续智能识别模型，并且本发明将UWB技术和激光探测技术结合，并利用TDOA算法进行计算，有利于实现对漂浮物的精准定位，进而有利于提高漂浮物的清理效率。

摘要： 本发明公开了一种基于rust‑IR2Graph与RGNN的rust漏洞检测方法。能够进一步提高rust漏洞检测的准确率。具体是通过rust‑IR2Graph进行rust‑IR转图操作，极大地丰富了代码的语法语义信息。使用创新的RGNN网络对图中各种类型的边有效建模，学习代码的结构特点，生成对应的特征向量以完成启代码漏洞检测任务。rust‑IR2Graph首先将标准化的rust程序进行解析，获取rust Mid‑Level IR，解析IR里的所有token和block，提取对应的数据流和控制流(有源码)，在指令级别进行转图操作，同时使用Word2Vec算法为节点生成特征向量。之后进入RGNN网络，该网络充分考虑到图中边的信息，通过邻域聚合学习待更新节点的邻域信息，从而自动学习代码的高维抽象特征用于代码漏洞检测任务。本发明在现有rust漏洞数据集上检测漏洞准确率达到90.7％。

摘要： 本发明公开了一种基于Graph OLAM的稀疏社交网络推荐系统，包括：用户界面模块、RUGOLAM算法引擎模块和后端数据库模块，其中，用户界面模块用于提供面向用户的界面，收集用户个性化需求；RUGOLAM算法引擎模块用于获取用户个性化需求，根据用户个性化需求生成分层多维聚集网络，计算分层多维聚集网络的节点间相似度，根据节点间相似度列出个性化推荐列表；后端数据库模块用于存储图形数据的属性和结构，并提供节点间相似度的计算支持。该系统化简针对稀疏网络的维度聚合操作执行过程，充分满足针对稀疏社交网络的推荐挖掘需求。

摘要： 本发明公开了一种基于Meta‑graph知识图谱表示的对话式音乐推荐方法，该方法包括基于Meta‑graph知识图谱的表示方法离线生成特征向量和基于Bandit算法的在线对话式推荐两部分。该方法实现对话场景下的基于Meta‑graph方法表示知识图谱的音乐推荐，即在人机对话的场景下，实时获取用户对音乐的偏好，并结合知识图谱来建模用户的长短期偏好，及时给予上下文感知的推荐结果，能够实现实时推荐，且对上下文信息、用户需求和反馈能良好处理。

摘要： 本发明涉及一种融合graph2vec、双层LSTM的车辆行程时间预测方法及系统，该方法包括：获取车辆历史出行轨迹数据，解析轨迹的相关特征信息，转换成键值对的形式存放，构建轨迹特征数据集；获取城市路网的路段唯一标识名与路段之间的邻接关系，构建拓扑空间数据集；对轨迹特征数据集中的特征因子与车辆行程时间进行相关性分析，选取强相关性的特征因子，并与拓扑空间数据集进行匹配，获取预测行程时间数据集；构建融合graph2vec、双层LSTM的车辆行程时间预测模型，将预测行程时间数据集输入到模型中，对车辆的行程时间进行预测，输出车辆行程时间。该方法及系统有利于更加精准的预测车辆行程时间，提高资源利用效率。

摘要： 本发明提出一种基于Nauru‑graph的大规模数据存储和读取方法和装置，其中，方法包括：获取数据ID值，检索数据ID值的可用存储节点，以生成数据的索引项ID值；以及，将数据的索引项ID值，存储于数据ID值的节点上，以对数据进行存储操作；步骤S2，计算数据的索引项ID值，通过索引项中数据ID找到目标节点，以对数据进行读取操作。本发明主要通过引入高对称强联通的拓扑结构改进数据存储时，数据分布均衡，系统稳定性和存储效率不能兼得的问题。同时给出了依据实际场景生成高对称强连通拓扑结构的方案。