一种基于自适应相似矩阵的谱聚类算法

为了消除在构建谱聚类算法的相似矩阵时,高斯核函数中尺度参数的波动影响,构建了一种自适应相似矩阵,并应用到谱聚类算法中。自适应相似矩阵中数据点间的距离度量采用测地距离算法,相距较近的两点间的距离近似于欧氏距离,相距较远的两点则先根据欧氏距离得到每个数据点的k个近邻点,然后累加近邻点的测地距离,由此得到每对数据点间的最短距离。两点间的局部密度用共享近邻的定义来表示,更好地刻画了数据集的本征结构。在5个人工数据集和国际通用UCI数据库中的5个真实数据集上进行实验。实验结果表明,所提算法的聚类准确率高于对比算法的准确率,对复杂分布数据有很强的自适应能力。研究成果为数据挖掘及机器学习提供了思路和方法。     

基于压缩感知理论和密度聚类算法的全回转推进器滚动轴承故障诊断

滚动轴承振动信号具有非线性、非平稳性等特性,且往往存在噪声干扰,信号特征难以提取,传统故障工况诊断方法受到应用限制。以全回转推进器为研究对象,提出一种基于压缩感知理论和密度聚类算法的滚动轴承故障诊断算法。首先,将振动信号分解,进行稀疏变换,得到稀疏域下的振动信号;其次,采用压缩感知理论进行信号重构,建立绝对值、偏斜度、方差以及裕度等特征指标;最后,采用密度聚类算法DBSCAN,将不同特征指标之间两两组合进行分析,成功区分了故障工况。为滚动轴承故障的无损化、智能化诊断提供了理论依据。     

小波聚类算法的分析与研究

聚类分析是数据挖掘领域中一个重要的研究课题。传统的聚类算法在在低维空间小数据集中能够有效地进行聚类,但在高维大数据集中,由于数据的稀疏性,存在大量孤立点等问题,聚类效果不甚理想。本文在传统聚类算法的基础上,着重对小波聚类算法进行了分析与研究。     

基于模拟退火算法优化谱聚类的有杆泵抽油井故障诊断

示功图是油田生产中分析井下生产工况的主要方法。随着计算机性能的不断提高,基于计算机的示功图分析方法越来越受到重视。传统的有监督学习方式受限于依赖主观经验进行训练样本分类。对此,本文采用无监督学习进行有杆泵抽油井的故障诊断。将示功图进行8方向链码的重画,提取其图形特征向量。然后由模拟退火算法优化谱聚类的方法进行故障的分类。将本文所提出方法用于一口抽油井的故障诊断中,验证了其有效性。     

基于改进谱聚类算法在图像分割中的应用

为解决谱聚类算法应用于图像分割时,相似矩阵内存占用较大甚至满溢以及后续计算量大的问题,利用Nystrom方法随机获取一部分样本点,根据样本点和样本点、样本点和非样本点2种相似关系近似表征所有像素点的相似性,得到原图像的近似相似矩阵。在构建上述所需2种相似关系的相似矩阵时,距离度量采用余弦函数。结果表明,采用近邻传播聚类算法代替k-means算法对得到的低维向量子空间聚类,克服了聚类过程对初始值的敏感性,得到的分割结果较稳定,4幅真实图片也验证了研究算法的优越性。改进的谱聚类算法为图像分割的稳定性研究提供了依据。     

一种优化K-means聚类的舌苔舌质分离方法

为了解决传统K-means聚类算法对舌图像中舌苔舌质分离效果不稳定的问题,提出了一种优化K-means聚类的舌苔舌质分离模型。模型综合RGB,HSV,L^*a^*b^* 3种颜色空间的特性,将单通道舌图像像素作为聚类样本集合,以一定的规则从集合中选取2个舌图像通道的像素作为聚类样本,并利用单通道舌图像分布直方图的峰值确定初始聚类中心点,最后通过聚类实现舌苔舌质分离。实验结果表明,选择用双通道舌图像作为聚类样本比用三通道或者单通道舌图像聚类效果更优,且提出的初始聚类中心点的确定方法相比于传统K-means聚类算法更准确、稳定地使舌苔舌质分离。所提方法可以很好地实现舌苔舌质分离,为中医舌诊客观化的研究提供了新的方法和思路,具有一定的实用价值。     

基于Carrot2聚类的垂直搜索引擎的研究与实现

给出了一个基于Nutch的垂直搜索引擎系统的实现,主要探讨了基于Lucene和Carrot2的信息检索与聚类的实现,并对分词、垂直信息采集等的实现进行了说明。     

基于模糊聚类的病险水库除险加固顺序研究

传统的病险水库除险加固顺序设置方法,在识别水库险情方面能力较弱,导致除险加固效果不佳.因此,提出基于模糊聚类的病险水库除险加固方法,并整理出加固顺序提高加固效率.在分析病险水库存在问题的基础上,对自然因素及人为因素进行聚类,设置模糊矩阵提取病险水库特征;并计算病险水库的风险识别参数,建立测损模型;计算病险水库对洪水的最...     

青风藤配方颗粒的制备及与传统汤剂的比较

为了研究青风藤配方颗粒制备工艺以及对青风藤配方颗粒与传统汤剂进行定性定量比较,运用正交试验方法,采用干法制粒,以出膏率、青藤碱含量为指标,优选青风藤配方颗粒制备工艺;采用薄层色谱法进行定性比较;以青藤碱为对照品,采用高效液相色谱法进行定量比较。青风藤配方颗粒最佳制备工艺如下:提取2次,加水量18倍(第1次加水量为10倍,第2次加水量为8倍),各煎煮1.0h。TLC结果显示,配方颗粒与传统汤剂化学成分的类型无明显差异。5批青风藤配方颗粒青藤碱含量分别为1.56%,1.58%,1.31%,1.20%和1.78%;5批青风藤传统汤剂青藤碱含量分别为0.94%,0.96%,0.85%,0.82%和1.02%。青风藤配方颗粒与传统汤剂中化学成分的类型基本相同,但青藤碱含量差异较大。     

The uniform distribution as a first practical approach to new product inventory management

It is common sense that the premises usually considered in inventory models have little applicability to new product inventory management. This paper develops a first practical approach to deal with this issue: the solution to the (Q, r) inventory model for uniform demand forecasts and lead-times. Based on the fact that the uniform distribution is defined by two parameters that are easy to estimate—maximum and minimum—this paper shows that such a premise may comprise a helpful and accurate decision support tool for managers until they begin to learn about the distribution characteristics of the demand during the lead-time.     

基于PSO的信息熵数据融合非均匀分簇路由算法

针对无线传感器网络分簇算法中能量分布不均衡导致的\"热区\"和簇头负载过重问题,提出了一种基于PSO算法优化簇头选举的非均匀分簇算法。在候选簇头选举和竞争半径计算过程中综合考虑节点动态能量、节点密度和节点距基站距离,将网络进行非均匀分簇,并引入PSO算法进行最终簇头选举。根据节点能量、节点密度和距基站距离确定簇间单跳多跳结合的路由规则,选取代价函数小的节点作为下一跳节点。基于节点信息熵确定融合阈值,进行簇内数据融合剔除冗余数据。仿真结果表明,改进算法的数据传输量比EEUC算法和UCRA算法分别提高了20%和10%,提升了数据的融合效率,有效延长了网络生命周期,簇头能量消耗得到均衡,减少了网络能量消耗,网络的整体性能显著优于其他对比算法。     

Firm Growth Under Sample Selection: Conditional <Emphasis>σ</Emphasis>-Convergence in Firm Size?

This paper tests the second, widely neglected implication of Gibrat’s law stating that the variance of firm size distribution increases over time. In contrast, learning models imply conditional σ-convergence in firm size. A unique data set of Austrian firms especially suited to account for sample selection effects is used to analyze the growth/size nexus of firms. The estimation results suggest that the predicted variance in firm size decreases only for firms of age 30 or below. For older age cohorts the hypothesis of a constant variance either cannot be rejected or increasing variances are found.     

Box-Behnken响应面法优化徐长卿配方颗粒挥发油-β-CD包合物制备工艺

为了解决徐长卿挥发油加油难及加入后的稳定性问题,采用Box-Behnken响应面法,优选出徐长卿配方颗粒挥发油-β-CD包合物的制备工艺。在单因素试验的基础上,采用3因素3水平Box-Behnken设计对徐长卿配方颗粒挥发油-β-CD包合工艺进行优选,以挥发油包合率及包合物收率为考察指标,考察挥发油与β-CD比例、包合温度、包合时间对结果的影响。结果表明:优选的包合工艺条件为挥发油与β-CD质量比为1∶8.34,包合温度为50.2℃,包合时间为2.17 h。在此条件下,挥发油包合率为72.19%,包合物收率为75.06%,综合评分为96.41,RSD值为1.1%。优选的包合工艺包合率高,工艺稳定可行,可为配方颗粒挥发油的工业化生产提供数据支持。     

Impact of personal and cultural factors on knowledge sharing in China

Knowledge sharing has been the focus of research for more than a decade and it is widely recognized that it can contribute to the success of an organisation. However, in comparison with other countries, relatively little work on this topic has been done in the Chinese context. Knowledge sharing is particularly interesting to study in the Chinese context at the individual level, given the unique social and cultural characteristics of this environment. In this paper, we develop a theoretical model to explain how personal factors would affect people’s intention to share their knowledge. The Theory of Reasoned Action and Social Exchange Theory are used in this paper, as are the time dimension of national culture, face, and guanxi. A survey methodology is used to test the model. Face and guanxi orientation both exert a significant effect on the intention to share knowledge. Theoretical and practical implications, as well as directions for future research, are discussed.

Getting consumers to eat more whole-grains: The role of policy, information, and food manufacturers

The 2005 Dietary Guidelines were unique because they offered quantitative recommendations for consumption of whole-grains. This case study examines the hypothesis that the changed recommendations were responsible for the recent increase in retail sales and consumption of whole-grain food products. We find that release of the Dietary Guidelines and related media attention did increase availability and sales of whole-grain foods. A large impact on consumption occurred through reformulation of existing products, induced by competition among food suppliers. This study reveals the key role product reformulation plays in inter-firm competition and in realization of dietary changes recommended by public policy.     

Gibrat's Law: Are the Services Different?

Several noted surveys on intra-industry dynamics have reached the conclusion from a large body of evidence that Gibrat's Law does not hold. However, almost all of these studies have been based on manufacturing or large scale services such as banking and insurance industries. There are compelling reasons to doubt whether these findings hold for small scale services such as the hospitality industries. In this paper we examine whether the basic tenet underlying Gibrat's Law– that growth rates are independent of firm size – can be rejected for the services as it has been for manufacturing. Based on a large sample of Dutch firms in the hospitality industries the evidence suggests that in most cases growth rates are independent of firm size. Validation of Gibrat's Law in some sub-sectors of the small scale services suggests that the dynamics of industrial organization for services may not simply mirror that for manufacturing. The present paper includes a survey of nearly 60 empirical studies on firm growth rates.