分析OBE模式要求,从大数据的背景、知识体系出发,探讨大数据专业的建设方法,大数据教育的专业定位、教学规划、课程设置、人才培养、教学方法等问题,给出相应的解决思路.

在移动自组织网络中节点自私性难辨,而自私节点不尽全力转发数据会降低网络的可靠性.基于此,将联合博弈论建立在动态贝叶斯框架下,形成了一种基于贝叶斯联合博弈的数据传输算法.算法首先根据各节点的信念更新值发现不良节点;然后根据不良节点预估网络环境,计算各个预估环境的联合信念概率并归一化为联合体存在概率,计算各联合体下各节点的安全效益值,得到所对应的合约;接着运用先验中期拒绝找到贝叶斯稳态合约;最后通过设置安全容量计算权值的方式惩处不良节点,保证联合体内各节点安全容量和收益最大化.仿真结果表明,与其他两种算法相比,所提算法不仅有较少的路由延迟和控制开销比,还有较高的数据包投递率,有效降低了不良节点对网络的不利影响.

谱聚类是当今机器学习领域的研究热点,大多数算法用于图像分割.由于谱聚类能够刻画数据在低维空间内的主要特性,因此分析了谱聚类表示特征的原理,构造了一种面向图像子块的非线性局部特征,提出了相应的特征提取算法,用于刻画SAR图像分块的性质.这些特征由谱聚类产生的若干特征值构成的向量组成,然后经过傅里叶变换得到,因而具有平移不变性.在计算的过程中,可以采用Nystr(o)m等方法解决谱聚类中矩阵不可逆问题.为了避免减弱局部特性差异,在子块相似性计算中采用了明氏距离.实验验证了所提特征的有效性.

Spectral clustering is able to find the nonlinear low-rank structure of data, and it is widely applied to pattern recognition. Besides,spectral clustering has some internal relations with graph models, manifold embedding and integral operator theory from the theoretical view. However, it is lack of systematically theoretical research in these aspects. The general model of spectral clustering is introduced from the latest research outcomes, that is, eigenfunctions learning of integral operators in reproducing kernel Hilbert space( RKHS) . Subsequently, the internal relations of spectral clustering with KPCA, kernel k-means, Laplacian eigenmap, manifold learning, and discriminant analysis are discussed. Then, some classical spectral clustering algorithms are introduced, such as NJW algorithm, Ncut, spectral clustering based on Nystr?m method, multiscale spectral clustering algorithm. At last, trends and possible difficulties in spectral clustering are summarized.

α stable distributions are the generalization forms of Gaussian distribution and Rayleigh distribution,capacity of accurately describing impulsive data and have widely applications in imaging and noise suppression.However,there are still lack of comprehensive introduction and comparative analysis in this aspect.Associating with current hot research points in speckle suppression and departing from MAP based filtering approaches,this paper introduces the definition,characteristic functions and properties ofαstable distributions and their special forms including symmetricalαdistribution,generalized Rayleigh distribution,Gauss distribution and some kinds of mixture distributions,and then analyzes some useful parameter estimation principles and methods.Subsequently,an experimental comparison was carried out and it verifies the superiority of MAP based filter.Besides,this paper introduces the newest applications in speckle suppression of these distributions.At the same time,this paper summaries the statistical characteristics of speckle and discusses some commonly used noise filtering algorithms,such as non-local mean approach.

随着网络高级持续威胁为主的犯罪活动逐渐增加,使得云计算环境下的高级持续威胁数据检测,具有高纬度、非线性等特征,导致传统基于PSO辨识树的高级持续威胁数据挖掘过程中,采用的数据主元特征以及关联特征存在显著的波动,无法获取准确的数据挖掘结果。提出了一种基于改进流形学习算法的云计算下高级持续威胁数据挖据模型,使用非线性流形学习算法,降低云计算下高级持续威胁数据向量特征的维数,通过特征提取模块对高级持续威胁数据进行预处理,采用改进经典流形学习算法,加大样本散布密集区域高级持续威胁数据间的距离,缩短样本散布稀疏区域样本间的距离,促使云计算下高级持续威胁数据样本库的整体分布均匀化,实现云计算环境下高级持续威胁数据的准确挖掘。实验结果说明,所提方法能够准确挖掘出云计算环境下的高级持续威胁数据,具有较高的挖掘效率和精度。

pectral clustering comes from operator theory and is able to efficiently decrease data dimension and classify data.However,current domestic researches pay more attention to applied algorithm design,and there is lack of theoretical outcome.In order to make up the shortcomings in theory,this paper systematically summaries the theory of spectral clustering,pays more attention to study the newest foreign operator theory research achievements.Besides,we briefly discuss and analyze some specific spectral clustering algorithms.We introduce and analyze the principle,convergence,current status of spectral clustering and its inherent association with manifold learning in terms of integral operator,spectral graph theory and manifold learning.At last,we suggest some possible directions.

Currently,spectral clustering is a state-of-art technique in image segmentation.However,the O(n3) complexity of spectral clustering its application in image segmentation.Based on the online multiscale compeittive learning,this paper proposes a new rapid spectral clustering algorithm for segmenting of images.This algorithm uses m(mn) constructed prototypes by online competitive learning to approximate the distribution of data and then groups prototypes by multiscale spectral clustering.With the approximate complexity O(mn+m2),our algorithm shows high performance and segmentation quality for large scale images.Our algorithm is tested on three data sets.On first data set,our algorithm shows correct grouping of data while NJW algorithm does not.Second,we compute the time consumption of our algorithm and NJW,and present the compression ratios in our algorithm.Results have shown our algorithm behavior better than NJW.At last the segmentation results on standard images of our algorithm take advantage of NJW and Kmeans algorithms and sampling based Nystrm grouping method.

Mixture Models(MMs) are a typical class of statistical models and have been applied to image processing in many situations, among which Gaussian MM (GMMs) are widely adopted. Main drawbacks of classical models involve that they need presetting the number of clusters, have not considered the influence of outliers. They will lead to unreasonable image segmentation results. This paper proposes the Self-Growing Regularized, GMMs(SGRGMMs), which generalizes the classical GMMs, for image segmentation. We compute the unknown parameters using the self-branching competitive leaning and a new generalized EM algorithm, Regularized EM(REM). We carried out experiments on the segmentation of some images and our approach can automatically determine the number of clusters and efficiently erase the influence of outliers.

Image segmentation is a key step for image processing and Gaussian Mixture Models(GMMs) are the common models for segmentation. The EM algorithm is usually used to estimete the parameters of GMMs, which is opt to get stuck at local minimum. In this paper we propose a new initialized shceme, multiscale online learning, for EM to aviod local minima and for GMMs to decide the optimal initial number of components. Experimental results have shown that this scheme can effectively improve the precision of segmentation compared to classical EM algorithm.

Existing publish/subscribe middleware are mostly optimized for static systems where users are fixed. In this paper, we present a novel protocol to support mobile clients for publish/subscribe middleware. We describe necessary steps of the protocol and how to solve the problem of message loss and duplication. The experiment shows that our protocol can effectively address the challenges raised by emerging mobile applications.

Clustering analysis is widely applied to engineering fields,such as biology sequence analysis,image segmentation,text analysis.Currently there have been many clustering methods and statistical learning based methods constitute a class of them.This paper started from FCM,introduced classical methods,such as potential and mountain functions,entropy method,and then analyzed their properties and applicability.Moreover,we also introduced the state-of-art clustering techniques,such as kernel clustering,spectral clustering and Gaussian mixture model based clustering,narrated the solving process and analyzed their properties,computation complexity.At last,this paper presented several research directions.

高斯混合模型(GMMs)是统计学习理论的基本模型,在可视媒体领域应用广泛.近些年来,随着可视媒体信息的增长和分析技术的深入,GMMs在(纹理)图像分割、视频分析、图像配准、聚类等领域有了进一步的发展.从GMMs的基本模型出发,从理论和应用的角度讨论和分析了GMMs的求解算法,包括EM算法、变化形式等,论述了GMMs的模型选择问题:在线学习和模型约简.在视觉应用领域,介绍了GMMs在图像分段、视频分析、图像配准、图像降噪等领域的扩展模型与方法,详细地阐述了一些最新的典型模型的原理与过程,如用于图像分段的空间约束GMMs、图像配准中的关联点漂移算法.最后,讨论了一些潜在的发展方向与存在的困难问题.

Spectral clustering receives wide attention in recent years since its efficiency in image segmentation and irregular data clustering. However, the applications of it in large scale data processing, such as web data categorization and image segmentation, are greatly restricted because of its O(n3) computational complexity. To address this problem, we propose a new effective scheme to greatly decrease the complexity while keep the clustering quality. The scheme adopts the Self-Organization Map(SOM) to encode the original data and then groups the obtained prototypes using multiscale spectral clustering proposed by us. We analyze and compare the performance of our approach with NJW and find that ours has less time consumption. Furthermore, we carry out an experiment on color image segmentation and results show that our approach behaves better than Kmeans algorithm.

An error concealment method based on Gaussian Mixture Model was proposed to solve the transmission errors of video streaming,and a detailed analysis,demonstration and research were conducted.With the temporal and spatial information adjacent to the lost blocks,an estimation was made of the lost pixel blocks with a minimum mean square error.When some video data were lost,an estimation of the lost pixel blocks with a minimum mean square error based on GMM was proposed.If the temporal information around the lost blocks was also missing,repeated estimation was used to solve the problem.The GMM increased the performance of PSNR compared to previously proposed methods of spatiotemporal error concealment,and the result was valid for a wide range of stationary loss probabilities.Simulations proved that the error concealment method based on GMM was better in enhancing and improving the subjective and objective equality of the video.

管道通信最能体现Linux平台的特色。分析了Linux平台下管道通信的实现机制,探讨了无名管道和有名管道的工作方式,并给出了相应的创建和使用的方法,同时指出了管道存在的不足。

Virtual reality is an important component of computer simulation,and plays an important role in the domain of industrial designing,commercial manufacture,virtual architecture,virtual combating environment,television advertising production,entertainment and so on.OpenGL,an open graphics language,has powerful ability in three-dimensional Render.Many software development platforms,such as Microsoft Visual C++,extend it as an important module.This paper introduces the definition,characteristics,principle of OpenGL,and then applies OpenGL to two examples from virtual reality.The pseudocode is also presented.

Using the methods of association analysis and clustering in the field of data mining, the paper focuses on the theories and methods of discovering user interests and points out the limitations of standard Web log. So it proposes a method of customized Web log in order to enhance the precision of user interests and preferences. The outcome of experiment shows that, by the method, Web log data hidden in the association rules as well as interests and preferences of similar users can be found, the precision of filtering user interest can be improved at the same time.

This paper proposes an online competitive learning algorithm, briefly denoted as KACL (Kernel Averaging Competitive Learning), using kernel functions and quadtree structure for clustering analysis and remotely sensed image segmentation. Initially, KACL constructs a quadtree with a pre-specified scale from online input data and then locates all clusters by moving or self-splitting the nodes of the quadtree. The complexity of the quadtree is decided by the distribution of data. In the learning rule of KACL, we use the local means of vectors but not single vector so as to avoid the movement of learning prototypes among different clusters. We give the mathematical properties of KACL and present the proof of its convergence. KACL avoids the dead node problem and presetting of the number of clusters. Two experiments are separately carried out on Gaussian mixture data sets and remote sensing images and the results have shown good performance.

With the expansion of manufacturing information services, visualization service in enterprise resource planning has recently become a hot topic. In this paper, we propose a new approach for job scheduling visualization service and a model of visualization service constructed from the share of production data and job scheduling information. Moreover, the optimization model of visualization service is given. At last, we discuss the optimization algorithm for this model.