聚类

基本解释 (translations)

clustering

与聚类相关的网络例句 [注：此内容来源于网络，仅供参考]

This paper proposes an algorithm of division and multi-level clustering with multi-strategy optimization,which bases on study of today's mature algorithms.

本文在研究已有成熟算法的基础上，提出了基于分治多层聚类的话题发现算法，其核心思想是把全部数据分割成具有一定相关性的分组，对各个分组分别进行聚类，得到各个分组内部的话题，然后对所有的微类再进行聚类，得到最终的话题，在聚类的过程中采用多种策略进行优化，以保证聚类的效果。

The main contribution of this thesis is that we propose a collaborative filtering recommendation algorithm based on double clustering. This approach first respectively clusters resources and users by the users rating on items, then makes a collaborative filtering recommendation based on the clustering result. The new algorithm can shorten on-line recommendation time. Then, we apply the classical formula of cosine correlation to double clustering algorithm, leaving out the standardization operation. In the end, we implement an educational resource recommender system according to the need of the actual project and the result is positive.

本文的主要意义在于，首先运用聚类技术对用户和资源分别进行聚类，然后利用聚类结果进行协同过滤推荐，由于聚类部分离线周期进行，大大缩短了在线的推荐时间；然后将经典的余弦相似性计算公式运用到双重聚类算法中，省去规范化处理操作，减少运算量；最后结合实际的需要，实现一个资源推荐系统。

This approach not only inherits the advantages of absolute density based algorithms which can discover arbitrary shape clusters and are insensitive to noises , but also efficiently solves the following common problems: clustering results are very sensitive to the user-deflned parameters, reasonable parameters are hard to be determined, and high density clusters are contained fully in coterminous low density clusters. With this approach, incremental clustering can also be supported effectively by defining the affected sets and seed sets of the updating objects in this approach.

聚类算法的有效性问题主要表现在三个方面：其一，聚类算法大多要求用户输入一定的参数，例如希望产生的簇的数目，而这些参数通常难以确定，特别是针对高维空间中稀疏分布的实际应用数据集，用户几乎无法给出合适的算法参数，因此非专业用户需要与数据分析专家密切配合才能保证获得理想的聚类结果，导致算法的使用极为不便；其二，聚类结果对于输入的参数值过于敏感J，往往参数值的一些轻微变化却产生聚类结果的很大差异；其三，对于高维的实际应用数据集，其数据分布往往是稀疏的、杂乱的，很难为算法选择全局的参数进行准确的聚类分析，使得聚类的质量难以保证。

From the starting value selection method in the Fuzzy ISODATA algorithm, used the method of maximal matrix element to ascertain the number of classification, the theoretical analysis of repeated test, and finally, the improved fuzzy ISODATA algorithm is obtained. The algorithm reduced sensitivity to the starting value. The algorithm can highly effective clustering analyze and obtains a stable result, so it presents an efficient way of improving the contribution value of custom service.

研究中先对模糊ISODATA聚类演算法中，初始划分矩阵和分类数的确定，并使用最大矩阵元素法，求得最佳化演算法中其他参数值，最后在CRM系统实际验证和分析中，采用本研究改良之模糊ISODATA聚类演算法，对汽车销售公司实施客户模组分类，经实验证明所得到的聚类结果，可有效解决一般聚类方法受限於参数设定敏感度的困扰，并使聚类精确度提高，有效排除杂讯敏感的影响，使聚类效果大幅提升，可帮助行销人员做出预测，制订出针对客户差异化的行销策略，提高客户服务的贡献价值。

I present cluster schema for combinatorial problem with clustering, design a new crossover operator cluster crossover, study the survival probability of c-schema under c-crossover, apply it successfully to graph coloring, load balancing and maximum spanning tree problems: give measurements of diversity in GA, set up varying crossover probability algorithms to maintain diversity, obtain good results when apply it in avoiding premature convergence and multi-modal functions optimization; establish GAC to be suitable for non stationary function, it can explore the change of environment and find the best solution even for some"anomalous"function.

针对聚类组合优化问题提出了聚类模式，设计了一种新的交换操作：聚类交换并研究了聚类模式在聚类交换操作下的生存可能性，成功的将聚类交换思想和算子应用于图着色、负载平衡和最大生成树问题，佐证了类理论，表明类理论可以指导新的应用算法设计；给出了GA中多样性的度量标准和维护多样性的动态交换概率算法，用于避免过早收敛和多峰函数优化，取得了较好的结果；构造了维持种群多样性以适应动态环境的方法GAC，在某些现有方法无能为力的问题上，GAC能够探测到环境变化，找到当前最优解。

A new hybrid clustering algorithm in phase of high-efficiency and good quality was put forward on the foundation of synthetically analyzing K-means clustering algorithm based on partition and agglomerative clustering algorithm based on hierarchy, and consulting some improved hybrid clustering algorithms.

在综合分析基于划分的K均值聚类算法和基于层次的凝聚聚类算法的基础上，借鉴各种混合聚类方法，提出了一种执行效率更高和聚类质量更好的分阶段混合聚类算法。

Under the guidance of DPRIF integrating strategy, we set up an integreted SOFM-SVM model. Then we analyzed and optimized the model from the aspects of operating mechanism, data interface and function expandedness. The PCA method was introduced to reduce dimensionality and extract features, then to strengthen the clustering explanation; defined a CMI index, to ascertain the most effective or the best clustering number; A new Anti-NO algorithm was proposed to recognize and to filter the suspecious data in the sample; The medium result of SVM model was used to extract the borderline datas between two classified groups. This research compensated for the achievements of data recognition including pattern data, noises data and borderline data.

在DPRIF整合策略指导下构建了一个整合的SOFM-SVM模型，对该模型从运作机制、数据接口、功能扩展几个方面进行分析和改进：引入PCA方法进行数据降维和特征提取，用以加强聚类解释；结合统计聚类中的聚类误差概念定义一个聚类数有效性指标，利用SOFM算法中间结果进行指标求解，以筛选出有效或最佳聚类数；提出一种新的噪声识别算法用于对样本中的异常数据进行甄别和过滤；利用SVM模型的中间结果提取分类边界数据；进一步充实了包括模式类、噪声集和边界在内的模式识别成果。

Compared wth traditional static clustering,dynamic automatic clustering has the following characteristics:real-time dynamicness,limited number of documents to be clustered every time,localness of document data to be used for clustering and randomness of the distribution of resources to be clustered in the whole resource set.

与传统静态聚类系统相比，动态自动聚类系统有以下特点：聚类是动态进行的，它是在检索结果返回的基础上进行的实时操作；每次聚类的文献对象数量有限；用来作为聚类依据的文献数据只是文献的局部；参与聚类的资源在整个资源集合中的分布是随机的。

The definition of the clustering and the algorithms in the clustering is introduced. We introduce the present situation of the clustering in time series, and now there are two kinds of clustering algorithm in time series, one is Adaptive Resonance Theory and their improvement algorithms; the other is Self-Organizing Feature Map and their improvement algorithms.

对聚类分析的概念作了简要介绍，讨论了现有的聚类分析中常用的方法以及时间序列的聚类分析的一些算法，当前聚类用于时间序列的符号化主要有两类，第一类是基于竞争学习模型的方法及其改进算法，第二类是自组织特征映射及其改进算法，本文对这两类算法分别作了探讨。

Researches on kernel clustering algorithms. After combining K-means clustering algorithm and the theory of kernel-based learning algorithms, we propose a fast kernel K-means clustering method which is based on CPD kernel. The experiment results indicate that the clustering effect of the algorithm is better than that of K-means algorithm, the clustering speed of the algorithm is also fast than that of K-means algorithm.

核聚类算法的研究：探讨了K-均值聚类算法，通过将核学习理论与K-均值聚类算法结合，提出了一种基于CPD核函数的快速核K-均值聚类算法，并将该算法与基于Mercer核的核聚类算法进行了比较，实验结果显示，我们的方法不仅比K-均值聚类算法的聚类效果好，而且聚类速度快。

与聚类相关的网络解释 [注：此内容来源于网络，仅供参考]

cluster set：聚类集

聚类搜索 cluster seeking | 聚类集 cluster set | 组开关 cluster switch

Cluster analysis：聚类分析

聚类分析(Cluster analysis)是定量研究分类问题的多元统计方法,属于一种探索性的分析,能够在没有先验知识的情况下将一批样本数据按照其性质上的密切程度自动分类,是非常适合于在不对类别特征做先验假设的情况下对国有商业银行和股份制银行进行重新分类.

Cluster analysis：聚类

我们以Cummins标准、CDA标准得到民族群体的肤纹结果,以自编的程序和聚类(cluster analysis)和主成分(principal component, PC)为分析的技术平台,做了综合性研究.

Quick Cluster：快速聚类

Quartile,四分位数 | Quick Cluster,快速聚类 | Radix sort,基数排序

clustering：聚类

同时还采取了一个称之为"聚类"(clustering)的新技术来彻底杜绝作弊现象. 另一家著名搜索引擎公司Excite更干脆,对网站标引时忽略Meta标签,以防止作弊. 但该做法对其他作弊现象,如大量重复关键词等则无能为力. 而且这种以噎废食的做法也不值得提倡.

clustering analysis：聚类分析法

聚类分析法:clustering analyzing | 聚类分析法:clustering analysis | 图像聚类:Image Clustering

hierarchical clustering：系统聚类

聚类算法:clustering method | 系统聚类:hierarchical clustering | 聚类分析法:clustering analyzing

Unsupervised Clustering：无监督聚类

无导师聚类:Unsupervised clustering | 无监督聚类:unsupervised clustering | 无监督异常检测:unsupervised anomaly detection

Unsupervised Clustering：无导师聚类

非监督分类:unsupervised classification | 无导师聚类:Unsupervised clustering | 无监督聚类:unsupervised clustering

hierarchical agglomerative method：集结层序聚类分析法

hidden observer phenomenon 隐蔽观察者现象 | hierarchical agglomerative method 集结层序聚类分析法 | hierarchical cluster analysis 层序聚类分析