有类间距离因素聚类结果的比较分析

本文对于有类间距离因素聚类结果的比较，提出了类结构的空间描述方法和比较相似度的度量指标──夹角余弦，并推导出它的一些性质．最后，用蒙特卡洛模拟的结果阐明用夹角余弦作为聚类结果的相似性度量指标是合理的．     

二维有序样本的有约束系统聚类

二维有序样本进行聚类必须满足两个要求：（１）类内各单元的相似性和类间的差异性；（２）各单元在位置上的有序性和类内的连通性。根据这些要求，将各单元观测指标间的距离矩阵作为聚类的指示矩将各单元之间的区位联系矩阵作为聚类的约束矩阵，在约束矩阵给出的约束条件之下，以类间单元指标的最大距离作为类间相似性指标，在指示矩阵中通过逐步聚并而将全部单元合并归类，即可得出满足要求的样本分类。     

中国区域创新能力静态分析——基于自适应赋权主成分聚类模型

本文在对经典聚类模型和现有改进聚类模型优点与不足剖析的基础上,通过定义客观加权主成分距离为分类统计量,提出了一种自适应赋权的主成分聚类模型。与现有同类方法相比,新模型克服了指标之间的高度共线性,能够对指标重要性的客观差异进行自适应赋权,每一步都有充分的理论保证其必要性、合理性.应用加权主成分聚类对中国区域创新能力进行集团划分,分类结果的可解释性明显提高,统计检验效果显著,所得的结论对了解和推动中国区域创新能力发展具有借鉴意义。     

数据的多流形结构分析

建立了基于ALM和权值的LRR聚类改进模型,对高维数据进行分析,将其分为两个独立的子空间,并与传统k-means聚类模型进行对比,采用评价指标模型对聚类结果进行评价分析.提出的LRR聚类改进模型在正则项引入了权重系数w,可以更好地将扰动分开,求解结果及评价指标均有效地验证了其稳定性、精确度等性能均有所提升.建立了SMMC改进模型,对机器工件外部边缘轮廓进行分类.从求解结果可看出该模型非常适合用于处理混合多流形聚类问题,对于比较复杂的曲线有着很好的分类性能.按照数据预处理、数据建模分析、模型结果评价步骤,通过使用谱聚类分析和多流形学习方法,对所给出的高维数据进行分析和处理,并通过评价模型得出相应的评价指标,对数据的多流形结构进行了深入的研究和探讨.     

基于指标信息量的灰色可能度聚类模型研究

针对灰色可能度聚类方法中指标权重确定问题,利用改进的CRITIC法确定指标权重,构建了改进的灰色可能度聚类模型.对于传统的CRITIC法,运用变异系数和点距离差值代替传统CRITIC法中的标准差和相关系数,以此代表指标间对比强度和冲突性并基于此提出了新型CRITIC法,解决了标准差不能有效反映指标变异程度,避免当指标数量较少时,相关系数波动性较大的问题.改进的灰色聚类模型通过各指标间信息量水平的差异从而确定指标的客观权重,有效体现了指标信息的重要作用,使权重的确定更加科学,最后运用算例验证改进灰色聚类模型的有效性.     

系统聚类分析中应注意的两类问题

给出了选用九种相似性度量，用最短距离法聚类，结果互不相同的一个有趣的例子。对该例，用欧氏距离求出距离矩阵后，除用最短距离法聚类结果唯一外，用最长距离法、重心法、类平均法、离差平方和法聚类，结果均不唯一。     

基于自适应权重的函数型数据聚类方法研究

基于有限维离散数据的传统聚类分析并不能直接用于函数型数据的分类挖掘。本文针对函数型数据的稀疏性和无穷维特殊性展开讨论,在综合剖析现有函数型聚类方法优势与不足的基础上,依据聚类指标的信息量差异重构加权主成分距离为函数相似性测度,提出了一种函数型数据的自适应权重聚类分析。相对同类函数型聚类算法,新方法的核心优势在于:(1)自适应赋权的距离函数体现了聚类指标分类效率的差异,并且有充分的理论基础保证其必要性和客观合理性;(2)基于有限维离散数据的聚类实现了无限维连续函数的聚类,能够显著降低计算成本。实证检验表明,新方法的分类正确率明显提高,能够有效解决传统聚类算法极端情形下的失效问题,有着复杂函数型数据分类问题下的灵活性和普遍适用性。     

一个新的模糊聚类有效性指标

提出了一个判别模糊聚类中聚类数有效性的新指标.首先利用FCM算法对数据集进行模糊聚类,通过隶属度矩阵和聚类中心构建加权二分网络.然后通过改进加权二分网络的模函数,定义一个新的聚类有效性指标.为了检验该有效性指标的性能,选取了三个常见的有效性指标在十五个数据集上进行了对比.实验结果表明,该有效性指标具有较好的性能.     

基于模糊聚类的属性匹配算法

提出了一种基于模糊聚类的属性匹配算法。该算法采用能综合反映属性名称相似性和语义相似性的模糊相似关系,提高了属性匹配的准确率;以等价闭包法对相似属性进行模糊聚类,得到多层次属性分类结果,更客观真实地反映了属性匹配的模糊性;同时,属性匹配过程中不需要设置匹配参数,避免了人为造成的误差。     

典型聚类分析

在医学科研与其它科研领域中,聚类分析方法已有广泛的应用.聚类分析包括两个内容:样品聚类与指标聚类,本文试图充实后一个内容.指标聚类的依据是相似系数,常用的相似系数是简单相关系数[1],从而导出了指标聚类的最大相关系数法和最小相关系数法.最大相关系数与最小相关系数均不能很好地反映两类指标间的相关关系,因此简单相关系数作为相似系数有一定的局限性,有可能使指标聚类结果无法解释.典型相关系数是简单相关系数的直接推广,它能较好地描述两类指标间的相关关系,自然可以用典型相关系数作为指标聚类的相似系数. 一、典型相关系数[2’3]…     

The Existence of Partitioned Generalized Balanced Tournament Designs with Block Size 3

A generalized balanced tournament design, GBTD(n, k), defined on a kn-set V, is an arrangement of the blocks of a (kn, k, k – 1)-BIBD defined on V into an n × (kn – 1) array such that (1) every element of V is contained in precisely one cell of each column, and (2) every element of V is contained in at most k cells of each row. Suppose we can partition the columns of a GBTD(n, k) into k + 1 sets B1, B2,..., Bk + 1 where |Bi| = n for i = 1, 2,..., k – 2, |Bi| = n–1 for i = k – 1, k and |Bk+1| = 1 such that (1) every element of V occurs precisely once in each row and column of Bi for i = 1, 2,..., k – 2, and (2) every element of V occurs precisely once in each row and column of Bi Bk+1 for i = k – 1 and i = k. Then the GBTD(n, k) is called partitioned and we denote the design by PGBTD(n, k). The spectrum of GBTD(n, 3) has been completely determined. In this paper, we determine the spectrum of PGBTD(n,3) with, at present, a fairly small number of exceptions for n. This result is then used to establish the existence of a class of Kirkman squares in diagonal form.     

等式约束优化的组合信赖域与拟牛顿算法

本文对等式约束问题提出了一个种组合信赖域与拟牛顿算法。该算法的特点是若Lagrangian函数的近似Ｈessian阵在等式约束Jacobi阵的零空间正定的,则选择拟牛顿算法,否则用信赖域算法,在通常信赖域算法的收敛假设下,该文证明了组合算法的全局收敛性。     

一类推广的拟牛顿多进程异步并行方法

本文通过近似雅可比矩阵Bk代替雅可比矩阵F′(xk),运用多进程异步并行方法求解非线性方程组。该方法在保持解的精度的情况下,缩短了运行时间和迭代步数。文中给出了算法收敛性的证明及八个非线性方程组的数值测试结果,表明该算法是可行的和快速的。     

微分方程组(dX)/(dt)=AX+e~(αt) sum from k=0 to m(B_kt~k)特解结构定理及应用

对于常系数非齐线性微分方程组(dX)/(dt)=AX+F(t),当强迫项F(t)=e~(at) sum from k=0 to m(B_kt~k)(这里Bk=(b1k,b2k,…,bnk)T∈Rn),给出了微分方程组(dt)/(dX)=AX+F(t)特解(t)的结构定理和计算方法,使求特解-X(t)的积分运算转化为简单的代数运算.解决了计算机特解(t)的计算问题.     

采用不定Dogleg路径的非单调自适应信赖域算法解无约束极小化(英文)

In this paper, we combine the nonmonotone and adaptive techniques with trust region method for unconstrained minimization problems. We set a new ratio of the actual descent and predicted descent. Then, instead of the monotone sequence, the nonmonotone sequence of function values are employed. With the adaptive technique, the radius of trust region △k can be adjusted automatically to improve the efficiency of trust region methods. By means of the Bunch-Parlett factorization, we construct a method with indefinite dogleg path for solving the trust region subproblem which can handle the indefinite approximate Hessian Bk. The convergence properties of the algorithm are established. Finally, detailed numerical results are reported to show that our algorithm is efficient.     

Kalman滤波的一种改型

对于n维状态x∈R~n与不完全观测y∈R~m(m相似文献   

Some Elements of Finite Order in K2Q

Let K2 be the Milnor functor and let Фn （x）∈ Q[X] be the n-th cyclotomic polynomial. Let Gn（Q） denote a subset consisting of elements of the form {a, Фn（a）}, where a ∈ Q^＊ and {, } denotes the Steinberg symbol in K2Q. J. Browkin proved that Gn（Q） is a subgroup of K2Q if n = 1,2, 3, 4 or 6 and conjectured that Gn（Q） is not a group for any other values of n. This conjecture was confirmed for n =2^T 3S or n = p^r, where p ≥ 5 is a prime number such that h（Q（ζp）） is not divisible by p. In this paper we confirm the conjecture for some n, where n is not of the above forms, more precisely, for n = 15, 21,33, 35, 60 or 105.     

A Generalized Machine-Scheduling Algorithm

It is proposed to treat a general machine-job-scheduling problem using a branch-and-bound method. Here the general problem is that in which the routing of any job through the machines is specified in advance but is independent of the routing of any other job. In addition there is no requirement for the job to visit all machines. Since any two operations to be performed on the same machine cannot be performed simultaneously, the set of all schedules can be divided into two subsets-one in which the pair of operations is performed in one order and the second in which the order is reversed. This division corresponds to branching in the method. For each of the new subsets formed by branching a lower bound on the duration of all schedules in the subset is calculated and the schedule with minimum duration is found by successive subdivision. An illustrative example is solved and the method is compared to other published general methods.     

The Place of Logic in Reasoning

Reasoning is a goal-oriented activity. The logical steps are at best the median part of a full reasoning: before them, a language has to be defined, and a model of the goal in this language has to be developed; after them, their result has to be checked in the real world with respect to the goal. Both the prior and the subsequent steps can be conducted rationally; none of them has a logical counterpart. Furthermore, Logic aims at prescribing what a correct reasoning is. But correct with respect to what? If the answer is: with respect to truth, the next question is whether the truth in everyday life, physics, economy, is the same as the truth that logicians have in mind. Resorting to Logic is justified only if an idealization in terms of true propositions in the logical sense is compatible with the goal. If such an idealization is legitimate, so is the use of classical Logic. If not, there is no authority forbidding to skew Logic in order to better reflect the nature of the reasoning required for the task.     

Robust energy planning in buildings with energy and comfort costs

Energy management in buildings is addressed in this paper. The energetic impact of buildings in the current energetic context is first depicted. Then the studied optimization problem is defined as the optimal management of production and consumption activities in houses. A scheduling problem is identified to adjust the energy consumption to both the energy cost and the inhabitant’s comfort. The available flexibilities of the services provided by domestic appliances are used to compute optimal energy plans. These flexibilities are associated to time windows or heating storage abilities. A constraints formulation of the energy allocation problem is given. A derived mixed linear program is used to solve this problem. The energy consumption in houses is very dependent to uncertain data such as weather forecasts and inhabitants’ activities. Parametric uncertainties are introduced in the home energy management problem in order to provide robust energy allocation. Robust linear programming is implemented. Event related uncertainties are also addressed through stochastic programming in order to take into account the inhabitant’s activities. A scenario based approach is implemented to face this robust optimization problem.