k-means cluster 中文意思是什麼
k-means cluster
解釋
類中心聚類-
First, the whole system was decomposed into several subsystems by adopting frizzy k - means cluster
首先，採用動態聚類方法，將整個系統分解為幾個子系統。 -
In this text, we first do some research on the genetic algorithm about clustering, discuss about the way of coding and the construction of fitness function, analyze the influence that different genetic manipulation do to the effect of cluster algorithm. then analyze and research on the way that select the initial value in the k - means algorithm, we propose a mix clustering algorithm to improve the k - means algorithm by using genetic algorithm. first we use k - learning genetic algorithm to identify the number of the clusters, then use the clustering result of the genetic clustering algorithm as the initial cluster center of k - means clustering. these two steps are finished based on small database which equably sampling from the whole database, now we have known the number of the clusters and initial cluster center, finally we use k - means algorithm to finish the clustering on the whole database. because genetic algorithm search for the best solution by simulating the process of evolution, the most distinct trait of the algorithm is connotative parallelism and the ability to take advantage of the global information, so the algorithm take on strong steadiness, avoid getting into the local
本文首先對聚類分析的遺傳演算法進行了研究，討論了聚類問題的編碼方式和適應度函數的構造方案與計算方法，分析了不同遺傳操作對聚類演算法的性能和聚類效果的影響意義。然後對k - means演算法中初值的選取方法進行了分析和研究，提出了一種基於遺傳演算法的k - means聚類改進（混合聚類演算法） ，在基於均勻采樣的小樣本集上用k值學習遺傳演算法確定聚類數k ，用遺傳聚類演算法的聚類結果作為k - means聚類的初始聚類中心，最後在已知初始聚類數和初始聚類中心的情況下用k - means演算法對完整數據集進行聚類。由於遺傳演算法是一種通過模擬自然進化過程搜索最優解的方法，其顯著特點是隱含并行性和對全局信息的有效利用的能力，所以新的改進演算法具有較強的穩健性，可避免陷入局部最優，大大提高聚類效果。 -
To the common requirement of invisibility of the two, the paper presents a information hiding strategy which first uses cluster analysis methods ( this paper uses k - means algorithm ) to classify the image to get the nature of the image, and then uses one embedding algorithm, accounting the improvement on the conceal effect using image ' s character
接著對信息隱藏技術的兩個重要分支隱寫術與數字水印技術，闡述其各自特點及異同點，針對二者在不可見性上的共同要求，從利於圖像自身的特性增強掩密效果的思路出發，提出先用聚類分析的方法（本文採用了k -均值演算法）對圖像的像素點進行分類以充分發掘圖像自然的內在特性，再結合嵌入演算法的信息隱藏策略。 -
Finally, according to the practical requirement of classification management to credit risk management, it uses k - means clustering method to cluster the evaluation result, and then get the credit ranks the small and middle enterprises belong to
最後，根據對信用風險管理應實行分級管理的實踐要求，利用k -平均聚類劃分法對信用風險評估結果進行聚類劃分，從而得到各中小企業所屬于的信用風險等級。 -
The paper puts forward the clustering algorithm includes : clustering based on grid and iterative, enhanced clustering algorithm base on density and k - medoids, enhanced k - means algorithm ( optimize chooseing consult _ points in iterative process ), enhanced clustering algorithm base on distance. they can overcome many limitations ( some traditional algorithms terminate in local optimization. many results of cluster are roundness, too many times in partition iterative process ), which are related to the static architecture of traditional model
在傳統聚類演算法的基礎上，結合我們科學數據挖掘的應用對象?分子動力學數據，提出了迭代網格聚類演算法， k -平均和基於密度結合的聚類演算法，迭代過程中優化選擇中心點的k -平均方法，以及改進型的基於距離的聚類演算法等模式識別方法，能夠解決傳統演算法帶來的諸多問題（比如一些傳統的聚類演算法常常收斂于局部最優，發現都模式都趨近於球形，劃分方法中迭代次數過多帶來的效率問題） 。 -
3. metal supported on ti02 ( 110 ) : calculation and simulation chapter 4, the properties of k, cu supported on the tio2 ( 110 ) surface have been studied by means of density functional theory, bare clusters models and embedded cluster model to using to obtain dft data and construct interatomic potential
3 .納米二氧化欽負載金屬體系的計算模擬研究我們對納米金紅石型tio :吸附k 、 cu金屬原子進行了dft研究，並模擬了金屬在納米金紅石型tio :表面的吸附行為，解釋了納米金屬簇在金紅石型tio :表面吸附的行為，預測了納米金屬團簇在表面生長的機理。 -
According to present situation of brand equity, for the first time, five important brand equity factors were extracted from brand features by applications of principal component analysis and factor analysis methods, they were brand status, customer - recognized value, brand image, brand creative abilities and brand executive abilities ; on the same time, five types of brand equity were divided with k - means cluster methods on the base of five brand factors, they were leading brand, matured or ripe brand, concrete brand, customer - based brand and creative brand. in order to extract brand equity strategy, correlation and linear regression analysis methods were used, as a result of analysis, four strategies were put forwarded including brand marketing strategy, marketing dividing strategy, marketing stretching strategy and marketing entrance time, applying nonparametric tests and duncan tests, five brand equities were also differed in many aspects
在品牌資產各組成要素中，應用主成分分析和因子分析方法，提取了五個品牌資產最重要的構成因子，首次提出品牌資產最重要的因子是品牌地位和顧客認知價值，其次為品牌形象、品牌創新能力和市場執行能力；根據品牌資產的構成因子，運用聚類分析法，對調查企業的品牌資產類型進行了分類，按照品牌構成屬性將企業分為領導型、成熟型、務實型、顧客導向型和創新型品牌企業；在對企業品牌策略分析基礎上，運用相關分析和線形回歸方法，求導形成品牌的重要策略因子，提出建立品牌資產最重要的策略因子是推廣策略，其次為市場分化策略、市場延伸策略和進入市場時機。 -
In samcluster system, the following cluster methods including hierarchical cluster analysis. k - means, and self - organizing map ( som ) and the feature selection methods based on coefficient of variation ( cv ) and simple t - test were integrated. to evaluate the performance of the samcluster system, the samcluster was applied to four expression datasets colon, leukemia72
在samcluster系統中，整合了下列聚類演算法：譜系聚類、 k -平均值聚類和自組圖聚類與變異系數計算和t -檢驗等基因變量選擇方法，並提出了一致的樣本分型概念，通過對四個基因表達譜的數據集colon 、 leukemia72 、 leukemia38和ovarian的測試，結果表明：誤判的樣本數分別為5 、 1 、 0和0個，因此，基因水平的樣本分型與樣本的臨床分型高度一致。 -
This paper introduces the development of data mining and the concepts and techniques about clustering will be discussed, and also mainly discusses the algorithm of cluster based on grid - density, then the algorithm will be applied to the system of insurance ? among the various algorithms of cluster put forward, they are usually based on the concepts of distance cluster o whether it is in the sense of traditional eculid distance such as " k - means " or others o these algorithms are usually inefficient when dealing with large data sets and data sets of high dimension and different kinds of attribute o further more, the number of clusters they can find usually depends on users " input 0 but this task is often a very tough one for the user0 at the same time, different inputs will have great effect on the veracity of the cluster ' s result 0 in this paper the algorithm of cluster based on grid - density will be discussed o it gives up the concepts of distance <, it can automatically find out all clusters in that subspaceo at the same time, it performs well when dealing with high dimensional data and has good scalability when the size of the data sets increases o
在以往提出的聚類演算法中，一般都是基於「距離（ distance ） 」聚類的概念。無論是傳統的歐氏幾何距離（ k - means ）演算法，還是其它意義上的距離演算法，這類演算法的缺點在於處理大數據集、高維數據集和不同類型屬性時往往不能奏效，而且，發現的聚類個數常常依賴于用戶指定的參數，但是，這往往對用戶來說是很難的，同時，不同參數往往會影響聚類結果的準確性。在本文里要討論的基於網格密度的聚類演算法，它拋棄了距離的概念，它的優點在於能夠自動發現存在聚類的最高維子空間；同時具有很好的處理高維數據和大數據集的數據表格的能力。
分享友人