模糊聚类在三维地震参数处理中的应用

三维地震参数是一个庞大的三维数据体，能从不同侧面反映储集层的特征。通过对三维地震参数的分析，可以评估油藏表征的参数（沉积相、构造、生长历史、流体饱和度等）。针对传统的聚类方法不能很好地综合考虑各个参数之间的相似程度，提出了用模糊C-均值聚类（fuzzy c-means clustering，简称FCM）对三维地震参数进行处理。从三维地震参数和油藏表征参数之间的关系出发，采用模糊C-均值聚类方法对三维地震参数进行处理，依此评估油藏表征的参数。测试结果表明，模糊C-均值聚类方法能够对三维地震参数进行较为准确的分类，并为储集层的研究提供了很好的依据。     

景像匹配区中稳定景物的分割方法研究

研究了景像匹配区中稳定景物的分割方法———强度和法及分形常数法 ,成功地分割了匹配区中的稳定景物 ,为评价区配区景物稳定性提供了依据 .其中分形常数法是分形理论应用在图像分割领域的一个新尝试     

模糊C-球壳聚类算法及其在血液细胞图像中的应用

文章从经典的模糊C均值算法开始通过改变其中相似性的度量形式，介绍了一种模糊C球壳聚类（FCSS）算法。在将该算法应用于细胞显微图像半径统计时，采用基于形态学的图像预处理措施，可以获得FCSS算法中有关原型模式的知识，加快收敛速度并避免随机初始化造成的局部极小问题。     

A Unified View on Clustering Binary Data

Clustering is the problem of identifying the distribution of patterns and intrinsic correlations in large data sets by partitioning the data points into similarity classes. This paper studies the problem of clustering binary data. Binary data have been occupying a special place in the domain of data analysis. A unified view of binary data clustering is presented by examining the connections among various clustering criteria. Experimental studies are conducted to empirically verify the relationships.     

红外警戒系统作战环境的计算机仿真

文中论述了红外警戒系统视景仿真研究的主要内容，给出了仿真软件流程。探讨了满足红外警戒系统全方位、多目标视景生成的实时性方法。     

利用U矩阵对SOM网络的处理

自组织神经网络的最大优点是能够保持原始数据的拓扑结构。但是当数据量很大的时候，自组织神经网络的神经元的数据也随之增大。因此为了更好地对数据进行分析，需要将自组织神经网络中相似的神经元进行分组，也就是聚类。在对SOM网络进行再次分析之前，为了减少“噪音”数据和孤立点对聚类结构的影响，用U矩阵的变型方法对自组织神经网络分析的结果进行预处理。     

一种基于Web文本聚类的用户兴趣发现模型的研究

研究了一种基于Web文本聚类的用户兴趣发现方法．他通过Web文档信息获取，文本的形式表示，以及Web文本聚类方法最终提取用户兴趣知识，并给出了一个设计模型。     

Studies of substrate specificities of lipases from different sources

Although lipases are widely applied in a wide variety of reactions, available information on the factors that are responsible for the substrate specificities of lipases from different sources is scarce. In this paper, nine lipase‐producing microorganism strains were isolated from oil‐containing samples. The properties of these enzymes, including pH optima, temperature optima, thermal stability, and pH stability, vary significantly with the different sources from which these lipases were isolated. The substrate specificities of the nine lipases, including fatty acid and positional specificities, were also studied. The fatty acid specificities of the nine lipases were observably different toward 15 substrates with different carbon chain lengths, different saturation degrees and different side chains. The lipases from S₃ Penicillium citrinum (PCL), MJ₁ Aspergillus niger (ANL), MJ₂ Aspergillus oryzae (AOL), YM Bacillus coughing (BCL), S₉ Geotrichum candidum (GCL), and S₁₁ Candida lypolytica (CLL) showed the strongest specificities to short‐chain esters, and the other lipases showed strong selectivity for medium‐ or long‐chain and branched esters. The positional specificities were examined by analyzing the hydrolytic products of triolein catalyzed by the nine lipases, using TLC. The lipases can be mainly classified into two groups by their specificities for the ester bond at position 2 of triglycerides; one group can selectively hydrolyze the ester bond at position 2 of the triglycerides, the other group cannot. All these nine lipases were divided into four groups by hierarchical clustering analysis on the basis of the results of fatty acid and positional specificities.     

Genetic algorithms for symbolic clustering

This paper introduces a novel methodology for clustering of symbolic objects by making use of Genetic Algorithms (GAs). GAs are a family of computational models inspired by evolution. These algorithms encode potential solutions to specific problems on simple chromosome-like data structures and apply recombination operators to these structures so as to preserve critical information. A new type of representation for chromosome structure is presented here along with a new method for mutation. The efficacy of the proposed method is examined by application to numeric data of known number of classes and also to assertion type of symbolic objects drawn from the domain of fat oil, microcomputers, microprocessors and botany. The validity of the clusters obtained is examined.     

Pseudoinverse in clustering problems

The clustering of vector observations of hyperplanes is studied. Different cases of correspondence distances are proposed and investigated, including the algebraic Jack Knife one. The efficiency, constructivity, and explicit form of formulas are provided by using the pseudoinverse technique including the pseudoinverse-perturbation theory. Results important for the application of pseudoinverse and related operators are presented. __________ Translated from Kibernetika i Sistemnyi Analiz, No. 4, pp. 73–92, July–August 2007.