一种基于K-Means局部最优性的高效聚类算法

K-Means聚类算法只能保证收敛到局部最优,从而导致聚类结果对初始代表点的选择非常敏感.许多研究工作都着力于降低这种敏感性.然而,K-Means的局部最优和结果敏感性却构成了K-MeanSCAN聚类算法的基础.K-MeanSCAN算法对数据集进行多次采样和K-Means预聚类以产生多组不同的聚类结果,来自不同聚类结果的子簇之间必然会存在交集.算法的核心思想是,利用这些交集构造出关于子簇的加权连通图,并根据连通性合并子簇.理论和实验证明,K-MeanScan算法可以在很大程度上提高聚类结果的质量和算法的效率.     

改进的基于局部模块度的社团划分算法

针对大多复杂网络社团划分算法不能快速发现最优节点加入社团的问题,提出一种利用节点亲密度的局部社团划分算法。引入节点亲密度的概念量化社团与邻居节点的关系,按照节点亲密度由大到小选择节点加入社团,最后以局部模块度为指标终止局部社团扩展。在真实网络和人工仿真网络进行实验,并与基于信息压缩的随机游走算法等4种典型社团划分算法相比较,所提算法划分结果的综合评价指标(F1score)和标准化互信息(NMI)均好于比较算法。实验研究表明,所提算法具有较好的时间效率和准确度,适用于大规模网络社团划分。     

无线传感器网络中基于Top-|K|查询分布式数据重构方法

针对分簇结构或多Sink节点的无线传感器网络应用场景,提出了一种基于Top-|K|查询的分布式数据重构方法.该方法包括分布式迭代硬阈值算法和基于双阈值的分布式Top-|K|查询算法两个部分.其中,管理节点和成员节点同时运行分布式迭代硬阈值算法,以分布式方式实现迭代硬阈值计算.同时,管理节点和成员节点运行基于双阈值的分布式Top-|K|查询算法,以分布式方式实现前一算法中查询绝对值最大的前K项元素和操作.实验结果表明,该方法的数据重构性能与现有方法无明显差异,同时能够有效地减少管理节点和成员节点之间的交互次数,并且降低网络中传输的数据量.     

基于最大生成树的社团划分算法

针对层次聚类算法存在复杂度高、准确度低等问题,提出了一种基于最大生成树的社团划分算法。该算法重新定义了节点间相似度,并利用最大生成树进行初始聚类,然后根据社团相似度合并局部社团得到最终划分结果。算法不仅降低了时间复杂度,而且在划分社团的准确度方面有所提高。将该方法在真实网络与人工网络上进行验证和比对,实验结果表明基于最大生成树的社团划分算法能够快速、准确地划分出网络中的社团结构。     

一种基于聚集系数的局部社团划分算法

社团划分算法是复杂网络研究中的一个热点问题.传统的复杂网络社团划分算法都必须获得全局网络的信息.随着网络规模不断增大,获得全局信息的难度随之增加;而在很多情况下只关心网络中某节点所在的局部社团.为了准确、快速地找到大规模复杂网络中的局部社团,提出了一种基于节点聚集系数性质的局部社团划分算法.该算法根据节点的连接频度,利用节点聚集系数的性质,从网络中某一待求节点开始,通过搜索邻居节点,划分该节点的社团结构.该算法只需要了解与待求节点相关的局部网络信息,在解决局部社团划分问题时其时间复杂度比传统的社团划分算法低.同时,该算法也可以应用于复杂网络全局社团结构的划分.利用该算法分别对Zachary空手道俱乐部网络和由Java开发工具包构成的软件网络图进行社团划分实验,并且分别对实验结果与对象网络的具体特征进行了对比分析.     

基于重要度贡献的无标度网络节点评估方法

针对无标度网络的节点重要度评估问题,通过分析节点的邻居数量与其邻居间的拓扑结构,得到节点的结构洞重要性指标,再融合相邻节点的K核重要性指标值来确定相邻节点间的重要度贡献,以此表征相邻节点的局部信息;在此基础上,再结合表征节点位置信息的节点自身的K核重要性,从而提出一种基于节点间重要度贡献关系来评估无标度网络的节点重要度的方法.该方法综合考虑了节点的结构洞特征和K核中心性特征来确定节点的重要度,同时兼顾到了网络的局部和全局重要性.理论分析表明,此方法的时间复杂度仅为o（n²）.与其他几种算法仿真对比的结果表明,该方法可行有效,拥有理想计算能力,适用无标度网络.     

基于节点相似性度量的社团结构划分方法

针对复杂网络结构划分过程复杂、准确性差的问题,定义了节点全局和局部相似性衡量指标,并构建节点的相似性矩阵,提出一种基于节点相似性度量的社团结构划分算法.其基本思路是将节点(或社团)按相似性合并条件划分到同一个社团中,如果合并后的节点(或社团)仍然满足相似性合并条件,则继续合并,直到所有节点都得到准确的社团划分.实验结果表明,所提算法能成功正确地划分出真实网络中的社团结构, 性能比标签传播算法(LPA)、GN(Girvan-Newman)、CNM(Clauset-Newman-Moore)等算法优秀,能有效提高结果的准确性和鲁棒性.     

基于相似度指标的社团划分算法

为快速准确地找到复杂网络的社团划分结果,引入相似度指标,提出一种改进的社团划分算法。将网络节点拓扑结构信息作为节点的多维属性,在不改变网络连边的情况下,使输入网络转换为节点多属性网络,并定义节点之间的混合相似度与社团相似度,运用层次聚类思想得到最终的社团划分结果。在真实网络、计算机生成网络上的实验结果表明,该算法能够发现明显的社团结构,并且具有较高的社团划分准确率。     

一种采用新型聚类方法的最佳类簇数确定算法

聚类分析是统计学、模式识别和机器学习等领域的研究热点.通过有效的聚类分析,数据集的内在结构与特征可以被很好地发掘出来.然而,无监督学习的特性使得当前已有的聚类方法依旧面临着聚类效果不稳定、无法对多种结构的数据集进行正确聚类等问题.针对这些问题,首先将K-means算法和层次聚类算法的聚类思想相结合,提出了一种混合聚类算法K-means-AHC;其次,采用拐点检测的思想,提出了一个基于平均综合度的新聚类有效性指标DAS（平均综合度之差,difference of average synthesis degree）,以此来评估K-means-AHC算法聚类结果的质量;最后,将K-means-AHC算法和DAS指标相结合,设计了一种寻找数据集最佳类簇数和最优划分的有效方法.实验将K-means-AHC算法用于测试多种结构的数据集,结果表明：该算法在不过多增加时间开销的同时,提高了聚类分析的准确性.与此同时,新的DAS指标在聚类结果的评价上要优于当前已有的常用聚类有效性指标.     

边赋权森林ω-路划分的O(n)算法

ω-路划分问题是路划分问题的一般化,它源于并行计算机系统、计算机网络与分布式控制系统等一类广播通信问题.设置最少的信息源节点,使得在指定的时间内将信息源节点所拥有的信息发送到其余节点,并且保证不同通信线路之间不得相交.从Hamilton路的NP-完全性不难看出,ω-路划分问题属于NP-完全问题.通过构造性证明技术,获得了边赋非负权路径、树和森林的ω-路划分问题的一些性质.分别提出了求解边赋非负权路径和边赋非负权树的ω-路划分问题的线性时间算法,讨论了算法的局部实现技术,详细地分析了这些算法的复杂度.以这两个算法为基础,提出了一个线性时间算法求解边赋非负权森林的ω-路划分问题.所提出的算法直观简明、操作容易,只需要较少的运行时间和较小的存储空间.     

A community detection algorithm based on Quasi-Laplacian centrality peaks clustering

Searching for key nodes in social networks and clustering communities are indispensable components in community detection methods. With the wide application demand of detecting community networks, more and more algorithms have been proposed. Laplacian centrality peaks clustering (LPC) is an efficient and simple algorithm which is proposed on the basis of density peaks clustering (DPC) to identify clusters without parameters and prior knowledge. Before LPC is widely applied in community detection algorithms, some shortcomings should be addressed. Firstly, LPC fails to search for key nodes in networks accurately because of the similarity calculation method. Secondly, it takes too much time for LPC to calculate the Laplacian centrality of each point. To address these issues, a community detection algorithm based on Quasi-Laplacian centrality peaks clustering (CD-QLPC) is proposed after studying the advantages of Quasi-Laplacian centrality which can replace density or Laplacian centrality to characterize the importance of nodes in networks. Quasi-Laplacian centrality is obtained by the degree of each node directly, which needs less time than Laplacian centrality. In addition, a trust-based function is utilized to obtain the similarity accurately. Moreover, a new modularity-based merging strategy is adopted to identify the optimal number of communities adaptively. Experimental results show that CD-QLPC outperforms many state-of-the-art methods on both real-world networks and synthetic networks.

     

Network representation learning based on community-aware and adaptive random walk for overlapping community detection

The Network representation learning methods based on random walk aim to learn a low-dimensional embedding vector for each node in a network by randomly traversing the network to capture the features of nodes and edges, which is beneficial to many downstream machine learning tasks such as community detection. Most of the existing random-walk-based network representation learning algorithms emphasize the neighborhood of nodes but ignore the communities they may form and apply the same random walk strategy to all nodes without distinguishing the characteristics of different nodes. In addition, it is time-consuming to determine the most suitable random walk parameters for a given network. In this paper, we propose a novel overlapping community detection algorithm based on network representation learning which integrates community information into embedding vectors to improve the cohesion degree of similar nodes in the embedding space. First, a node-centrality-based walk strategy is designed to determine the parameters of random walk automatically to avoid the time-consuming manual selection. Second, two community-aware random walk strategies for high and low degree nodes are developed to capture the characteristics of the community centers and boundaries. The experimental results on the synthesized and real-world datasets demonstrate the effectiveness and efficiency of our algorithm on overlapping community detection compared with the state-of-the-art algorithms

     

Clustering distributed data streams in peer-to-peer environments

This paper describes a technique for clustering homogeneously distributed data in a peer-to-peer environment like sensor networks. The proposed technique is based on the principles of the K-Means algorithm. It works in a localized asynchronous manner by communicating with the neighboring nodes. The paper offers extensive theoretical analysis of the algorithm that bounds the error in the distributed clustering process compared to the centralized approach that requires downloading all the observed data to a single site. Experimental results show that, in contrast to the case when all the data is transmitted to a central location for application of the conventional clustering algorithm, the communication cost (an important consideration in sensor networks which are typically equipped with limited battery power) of the proposed approach is significantly smaller. At the same time, the accuracy of the obtained centroids is high and the number of samples which are incorrectly labeled is also small.     

基于染色随机游走的可重叠社区发现

社区发现是一个基础性的且被广泛研究的问题。现有的社区发现方法大多聚焦于网络拓扑结构,然而随着真实网络中实体可用属性的激增,捕获图中结构和属性的丰富交互关系来进行社区发现变得尤为必要。据此面向属性图提出了一种基于染色随机游走的可重叠社区发现算法OCDC,该算法解决了传统的基于随机游走的社区发现算法利用结构转移矩阵造成社区发现效果不佳的问题。具体地,首先利用经典的初始种子策略选出网络中差异度较大的节点,在此基础上设计种子替换策略,挖掘网络中质量更佳的种子替换路径集合对初始种子集合进行替换;其次构建结构-属性交互节点转移矩阵并执行染色随机游走过程得到高质量种子节点的染色分布向量;最后基于融合结构和属性的并行电导值对社区进行扩展。在人工网络和现实网络上的实验表明,本文提出的算法能够准确地识别属性社区并显著优于基准算法。     

一种加权稠密子图社区发现算法

目前,针对复杂网络的社区发现算法大多仅根据网络的拓扑结构来确定社区,然而现实复杂网络中的边可能带有表示连接紧密程度或者可信度意义的权重,这些先验信息对社区发现的准确性至关重要.针对该问题,提出了基于加权稠密子图的重叠聚类算法（overlap community detection on weighted networks,简称OCDW）.首先,综合考虑网络拓扑结构及真实网络中边权重的影响,给出了一种网络中边的权重定义方法;进而给出种子节点选取方式和权重更新策略;最终得到聚类结果.OCDW算法在无权网络和加权网络都适用.通过与一些经典的社区发现算法在9个真实网络数据集上进行分析比较,结果表明算法OCDW在F度量、准确度、分离度、标准互信息、调整兰德系数、模块性及运行时间等方面均表现出较好的性能.     

A Semantic Searching Scheme in Heterogeneous Unstructured P2P Networks

Semantic-based searching in peer-to-peer (P2P) networks has drawn significant attention recently. A number of semantic searching schemes, such as GES proposed by Zhu Y et al., employ search models in Information Retrieval (IR). All these IR-based schemes use one vector to summarize semantic contents of all documents on a single node. For example, GES derives a node vector based on the IR model: VSM (Vector Space Model). A topology adaptation algorithm and a search protocol are then designed according to the similarity between node vectors of different nodes. Although the single semantic vector is suitable when the distribution of documents in each node is uniform, it may not be efficient when the distribution is diverse. When there are many categories of documents at each node, the node vector representation may be inaccurate. We extend the idea of GES and present a new class-based semantic searching scheme (CSS) specifically designed for unstructured P2P networks with heterogeneous single-node document collection. It makes use of a state-of-the-art data clustering algorithm, online spherical k-means clustering (OSKM), to cluster all documents on a node into several classes. Each class can be viewed as a virtual node. Virtual nodes are connected through virtual links. As a result, the class vector replaces the node vector and plays an important role in the class-based topology adaptation and search process. This makes CSS very efficient. Our simulation using the IR benchmark TREC collection demonstrates that CSS outperforms GES in terms of higher recall, higher precision, and lower search cost.     

A stable community detection approach for complex network based on density peak clustering and label propagation

Dividing a network into communities has great benefits in understanding the characteristics of the network. The label propagation algorithm (LPA) is a fast and convenient community detection algorithm. However, the community initialization of LPA does not take advantage of topological information of networks, and its robustness is poor. In this paper, we propose a stable community detection algorithm based on density peak clustering and label propagation (DS-LPA). First, the local density calculation method in density peak clustering algorithm is improved in finding the community center of the network, so as to build a suitable initial community, which can improve the quality of community partition. Then, the label update order is determined reasonably by computing the information transmission power of nodes, and the solutions for multiple candidate labels are provided, which greatly improved the robustness of the algorithm. DS-LPA is compared with other seven algorithms on the synthetic network and real-world networks. NMI, ARI, and modularity are used to evaluate these algorithms. It can be concluded that DS-LPA has a higher performance than most comparison algorithms on synthetic network with ten different mixed parameters by statistical testing. And DS-LPA can quickly calculate the best community partition on different sizes of real-world networks.

     

基于网络嵌入与局部合力的复杂网络社区划分算法

社区划分可以揭示复杂网络中的内在结构和行为动态特点,是当前的研究热点。文中提出了一种基于网络嵌入和局部合力的社区划分算法。该算法将网络的拓扑空间转化成欧氏空间,把网络节点转换成向量表示的数据点,首先基于重力模型和网络拓扑结构,提出局部合力和局部合力余弦中心性指标(Local Resultant Force Cosine Centrality,LFC),通过节点的LFC和节点间的距离来确定各个初始小社区的中心节点,然后将网络中其他的非中心节点划入与其最近的中心节点所在的初始小社区内,最后通过优化模块度的方法来合并初始小社区并找到最优的网络社区结构。在6个现实世界网络和可调参数人工网络上与6种知名社区划分方法进行比较,比较结果表明了新算法良好的社区划分的性能。