检索结果-内蒙古大学图书馆

A novel hierarchical clustering method in signed social network

Journal of Computational Information systems 2015年第21期11卷 7865-7872页

作者： Guo, Jingfeng Chen, Xiao Fan, Chaozhi Hu, Xinzhuan College of Information Science and Engineering Yanshan University Qinhuangdao China The Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Qinhuangdao China Qian'an College North China University of Science and Technology Qian'an China

Based on a new feature that macro network structure and micro game have mutual inuence on the formation of the community, to mining communities in signed network, a novel algorithm SNCGHC is proposed. Firstly, according to the signed attributes and the node degree, the Shapley Value is improved, to avoid ignoring the effect of external environment on the Shapley Value and reduce the calculation complexity. Secondly, considering the inuence of network density on the formation of the coalition (community), we find the calculation method of average coalition payoff and create the signed network coalitional game model. Thirdly, SNCGHC depends on average payoff maximization of coalition to achieve hierarchical clustering of signed network. Finally, experiments show that our method can detect better community structures as compared to other existing methods. Copyright © 2015 Binary Information Press.

关键词： Game theory

来源：评论

学校读者我要写书评

暂无评论

A novel approach to identify influential functions in complex software network based on complex network

引用

ICIC Express Letters 2016年第2期10卷 485-492页

作者： Huang, Guoyan Chen, Xiaojuan Wu, Hongfei Zhang, Peng Ren, Jiadong College of Information Science and Engineering No. 438 West Hebei Ave Qinhuangdao China Computer Virtual Technology and System Integration Laboratory of Hebei Province Yanshan University No. 438 West Hebei Ave Qinhuangdao China

Identifying influential nodes is an important issue in understanding the process of information diffusion in complex software networks. Researchers generally define functions as nodes, and relationship of function calls as edges to map software into complex network. This paper proposes a novel approach to identify the influential nodes in complex software network based on complex network. Firstly, we map the function and relationship of function calling or dependency to a directed weighted call network (DWCN). Secondly, we define NodeScore (NS) to evaluate the importance of nodes in network. It has higher influence when the node has lager value of NS. The NS of each node is calculated by two parts in our opinion, Direct Dependency Nodes (DDN) and Indirect Dependency Nodes (IDN). Thirdly, algorithm ComputeNodeScore is proposed to calculate NS of each node. Finally, we mine top-k nodes based on the NS. Experimental results of various versions about software Tar and cflow show the approach is effective for identifying the influential nodes. © 2016, ICIC Express Letters ICIC International.

关键词： Complex networks

来源：评论

学校读者我要写书评

暂无评论

Mining important nodes in directed-weighted software network based on semi-global structure information

引用

ICIC Express Letters 2015年第10期9卷 2859-2866页

作者： Ren, Jiadong Wu, Hongfei Yin, Tengteng Liu, Yang He, Haitao College of Information Science and Engineering Yanshan University No. 438 West Hebei Ave. Qinhuangdao China The Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Yanshan University No. 438 West Hebei Ave. Qinhuangdao China

Software’s control flow graph can be treated as a complex software network, and a tiny fraction of key function nodes has considerable influence on the stability, reliability and robustness of the network. Identifying the top-k key nodes accurately and effectively has become an important topic in software engineering. In this paper, directedweighted function dependency network (DWFDN) is adopted to represent the function and relationship of function calling or dependency. The weight on a direct edge indicates the frequency of function calling. Weighted semi-global structure information (the weight, the relative position of a function node and the topological structure among its neighbor nodes) is taken into consideration in our approach. From the perspective of accumulating and propagating vulnerabilities, key function callers and function callees are mined respectively. Case studies on real software date show the utility of the approach. © 2015, ICIC International.

关键词： Software reliability

来源：评论

学校读者我要写书评

暂无评论

An improved Wu-Huberman community discovery algorithm based on edge points

引用

ICIC Express Letters 2013年第11期7卷 2949-2955页

作者： Guo, Jingfeng Wang, Xiaolin Zuo, Mengfei Lv, Jiaguo College of Information Science and Engineering Yanshan University No. 438 Hebei Ave. Qinhuangdao 066004 China The Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Yanshan University No. 438 Hebei Ave. Qinhuangdao 066004 China

There are two critical questions in the Wu-Huberman algorithm: one is picking two poles and the other is setting the threshold. By further studying the two questions, we present a method that makes the discovery of communities more precise and the scope of application of the algorithm wider. The method picks the edge nodes of two communities as two poles and finds the largest gap using the core nodes of communities, which can discover communities of arbitrary size instead of roughly the same size. © 2013 ISSN 1881-803X.

关键词： Poles

来源：评论

学校读者我要写书评

暂无评论

Angle historgam model of directional relationship in 3D based on Double-projection and Rounded-subdivision

引用

Journal of Information and Computational Science 2014年第18期11卷 6471-6478页

作者： Kong, Dehan Liu, Yongshan Ma, Junlei Department of Information Science and Engineering Yanshan University Qinhuangdao China The Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Yanshan University Qinhuangdao China Data Industry Research Lab Qinhuangdao China

The formal model of spatial directional relations is one of the most important parts in spatial relation research. The most of models are based on Minimum Bounding Rectangle (MBR), and they are not compliant with the regular pattern of human cognition. In order to get a closer conclusion to human cognition on directional relationship, Angle Histogram model based on Double-projection and Rounded-subdivision (AHDPRS) is proposed in this paper. The model uses the maximum inscribed circles to find out the maximum parts of the object, and calculates the directional relationship between the centers of the circles. This model ignores the inessential details to ensure the result which will be closer to human cognition. The experiments show that this model is feasible.

关键词： Graphic methods

来源：评论

学校读者我要写书评

暂无评论

LoBa-Min-Min-SPA: Grid load balance using SPA

引用

Open Automation and Control systems Journal 2013年第1期5卷 87-95页

作者： Zhang, Zhongping Wen, Lijuan Wang, Zhiping College of Information Science and Engineering Yanshan University Qinhuangdao Hebei 066004 China Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Qinhuangdao Hebei 066004 China Foreign Language College of Dalian Jiaotong University Dalian Liaoning 116052 China

In the grid environment, there are a large number of grid resources scheduling algorithms. According to the existing Min-Min scheduling algorithm in uneven load, and low resource utilization rate, we put forward LoBa-Min-Min algorithm, which is based on load balance. This algorithm first used Min-Min algorithm preliminary scheduling, then according to the standard of reducing Makespan, the tasks on heavy-loaded resources would be assigned to resources that need less time to load balance, raise resource utilization rate, and achieve lesser completion time. We used benchmark of instance proposed by Braun et al. to prove feasibility and effectiveness of the algorithm. At Last, We introduced the SPA and gave the experimental results of Min-Min-SPA and LoBa-Min-Min-SPA. © Zhang et al.;Licensee Bentham Open.

关键词： Grid computing

来源：评论

学校读者我要写书评

暂无评论

An efficient Subspace Clustering Algorithm based on Attribute Relativity and DBSCAN

引用

ICIC Express Letters 2014年第6期8卷 1677-1682页

作者： Zeng, Qiang Bai, Yunyue He, Haitao Ren, Jiadong College of Information Science and Engineering Yanshan University No. 438 Hebei Ave. Qinhuangdao 066004 China The Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Yanshan University No. 438 Hebei Ave. Qinhuangdao 066004 China

In the high-dimensional data traditional clustering algorithms tend to break down because of the curse of dimensionality, high cost of time, etc. This paper proposes a novel algorithm AReSUBCLU, an Effective Subspace Clustering Algorithm based on Attribute Relativity and DBSCAN. Active attribute is defined to reduce the dimension of data, and a subspace search tree which is constructed by attribute correlation matrix is defined to generate subspaces. Firstly, this method uses DBSCAN with the technique of weighted threshold to cluster on every active attribution. Secondly, the correlation matrix is constructed by computing the covariance of attributes. And the subspace search tree is constructed by the matrix which is used to measure the relativity of attributions. The set of nodes in each root branch of the tree is the interesting subspaces that the algorithm discovers. Finally, the subspace clusters are obtained through merging every cluster of active dimension with the usage of similarity of cluster. Evaluation with synthetic and real-world datasets demonstrates that AReSUBCLU has a good clustering quality, and capability of discovering subspace clusters of arbitrary shapes. © 2014 ISSN 1881-803X.

关键词： Clustering algorithms

来源：评论

学校读者我要写书评

暂无评论

Point cloud coarse registration based on 3DSIFT keypoint detection

引用

Journal of Computational Information systems 2014年第23期10卷 10121-10128页

作者： Gu, Xiaoying Liu, Yongshan Department of Information Science and Engineering Yanshan University Qinhuangdao China The Key Laboratory for Computer Virtual Technology and System Integration of HeBei Province Yanshan University Qinhuangdao China Data Industry Research Lab Qinhuangdao China

Since the SIFT feature point extraction algorithm with scale changes, rotation transformation invariance, is widely used in image registration. In this paper, the SIFT algorithm is applied to three-dimensional point cloud coarse registration, the proposed 3DSIFT extraction algorithm is suitable for three-dimensional point cloud data, then point coordinates, curvature, the nearest neighbor distance mean and other information compose fourteen-dimensional vector to conduct correspondence match, use the interior point rate of Ransac to obtain optimal transformation, and finally transform the coordinates for source point clouds using the optimal transform, complete the point cloud data coarse registration. Experimental results show that our coarse registration algorithm can effectively extract feature points, and it is robust for the point cloud with noisy point, it can provide accurate and effective initial value for the precise registration such as ICP.

关键词： Metadata

来源：评论

学校读者我要写书评

暂无评论

Research on a denoising smoothing algorithm for 3D scattered point cloud

引用

ICIC Express Letters 2014年第9期8卷 2403-2409页

作者： Gu, Xiaoying Liu, Yongshan Wu, Qiong Department of Information Science and Engineering China The Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province Yanshan University No. 438 Hebei Ave. Qinhuangdao 066004 China Data Industry Research Lab Qinhuangdao 066004 China

In order to effectively eliminate outliers and noise points in 3D scattered point cloud, a denoising smoothing algorithm which is the combination of removing outliers algorithm and trilateral filter is proposed. This is accomplished by rejecting outliers, correcting the normal direction and moving noise points along the modified normal direction within local adaptive neighborhoods. First, the outliers are removed by the method of mathematical statistics, on this basis, estimating normal direction and curvature of point cloud by covariance analysis method. Then, correct the normal direction by combining trilateral filter weighting factor with adaptive optimal neighborhood function. Finally, calculate position offset of data points in the direction of normal amendment by trilateral filter weighting factor in order to smooth the point cloud. Experimental results show that the proposed method is simple and effective, successfully removing noise from point cloud while preserving inherent geometric features. © 2014 ISSN 1881-803X.

关键词： Bandpass filters

来源：评论

学校读者我要写书评

暂无评论

Reinforced Heterogeneous Graphlet Design for Knowledge Graph Representation Learning

SSRN

引用

SSRN 2023年

作者： Gong, Jibing Lin, Yuting Zhao, Yi Lin, Tianyu Fang, Xiaohan Feng, Xinchao Peng, Jiquan School of Information Science and Engineering Yanshan University Qinhuangdao066004 China The Key Laboratory for Computer Virtual Technology System Integration of Hebei Province Yanshan University Qinhuangdao066004 China

Knowledge Graphs have been practical tools to represent and integrate plentiful structural and semantic information in mainstream industrial scenarios. While promising, the heterogeneity and complexity of KGs pose a formidable obstacle, especially for graph representation learning. Most existing KG embedding models omit the dynamic high-order connectivity patterns, and heavily rely on the handcrafted patterns to handle the complex semantic relationships, thereby limiting their ability to adaptively capture nuanced and intricate relationships of knowledge graphs in different tasks. To fill this gap, we present ReHGD — a model designed for KGs that focuses on adaptively designing typed graphlets through a cooperative Multi-agent Reinforcement Learning algorithm. This approach can learn more discriminative graph representations tailored to specific downstream tasks. Specifically, ReHGD engages in the creation of typed graphlets through a two-stage process: it (1) establishes a reinforced chain design module to generate chains without predefined rules, and (2) employs a buffer-aware sampling technique to derive episodic chains from prior experiences. Subsequently, motifs are deduced by applying commute count and Hadamard product operations to the episodic chain-based subgraphs. Finally, ReHGD utilize chain and motif aggregations to learn graph representations. Experimental results on three real-world KGs show that ReHGD outperforms baselines on entity classification. © 2023, The Authors. All rights reserved.

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：