一种基于回归分析的空间划分方法

针对子空间连通性约束情况下空间对象的局部回归分析问题,本文提出基于最小拟合误差平方和准则的空间划分方法,从一个空间的初始划分开始,按照拟合误差平方和下降的原则调整子空间边界,获得新的空间划分和对应的回归模型,不断迭代直到准则函数收敛。     

基于子空间划分的模糊系统模型辨识

提出了基于子空间划分的模糊系统模型（SPFS），并给出一种针对SPSF的白适应模型辨识方法．应用遗传算法进行子空间划分方案的优化。降低了最大子空间的辨识误差，从而得到优化的模型辨识结果．理论分析和仿真计算证明了该模型的有效性．所提出的模型有助于缓解规则数爆炸问题．     

全最小二乘和全加权最小一乘准则下的空间直线拟合

讨论空间直线拟合问题,在使得点到空间直线距离平方和最小准则下,给出一种基于主成分分析的直线拟合方法,并且在使得空间点到直线加权距离和最小准则下,给出了拟合直线的优化方法.算法简单,在计算机上实现方便.最后,数值仿真验证了算法的有效性,并讨论了空间中圆周拟合问题.     

利用频域拟合冗余原理直接求取连续系统离散降阶模型

本文利用频域拟合给出直接从高阶连续模型求离散降阶模型的模型降阶方法．文中首次提出频域拟合冗余及关键方程组的概念，论证了根据关键点拟合原理进行一步法频域拟合降阶的优越住，给出智能化点拟合算法，有效地求解了一类关于模型参数空间的非线性误差准则的极小化问题．     

非均匀B样条曲面的自适应节点设置方法

针对非均匀B样条的节点设置问题,提出一种利用非均匀B样条曲面拟合离散数据的迭代算法,通过优化节点分布来改进拟合曲面的质量.该算法以带参数化的三角网格曲面为输入,在首次迭代中根据输入曲面的几何特征将其对应的参数域划分成若干个子区域,并使得每个子区域上累积的几何特征信息量近似相等,子区域的重心坐标将取为首次迭代的节点;在随后的迭代中,保证前次迭代生成的重心位置固定不变,并根据前次迭代得到的曲面拟合误差再次将区域划分成累积误差接近相等的子区域,新增加的子区域重心的坐标选为拟加入的节点.文中算法自适应地在曲面形状复杂或拟合误差大的区域引入更多的控制顶点,使得拟合曲面的质量得以逐步改进.实验结果表明,该算法快速有效,在拟合具有明显几何特征的输入数据时具有优势.     

基于误差预测的焦炉火道温度软测量模型

针对焦炉火道温度在线检测的问题,在分析焦炉火道温度特性的基础上,建立了一种基于误差预测的高精度焦炉火道温度软测量模型．首先分别建立了1元、2元和12元线性回归模型,对蓄顶温度和火道温度进行线性拟合;然后比较分析了三种回归子模型的特点．使用融合时间差分法的Elman神经网络,对线性回归模型中精度最高的12元模型的预测误差进行拟合和多步预测．采用专家经验将线性回归组合模型和融合时间差分法的Elman神经网络模型进行集成,最终获得了具有较高预测精度的焦炉火道温度软测量值．实际运行结果验证了该软测量模型的有效性．     

利用频域似合冗余原理直接求取连续系统离散降阶模型

本利用频域拟合给出直接从高阶连接模型求离散降阶模型的模型降阶方法。中首次提出频域拟合冗余及关键议程组的概念，论证了根据关键点拟合原理进行一步法频域拟合降阶的优越性，给出智能化点拟合算法有效地求解了一类关于模型参数空间的非线性误差准则的极小化问题。     

基于拟合的二维正态云模型

分析了现有的正向/逆向云生成器实现算法的不足之处,并在此基础上提出了基于拟合思想的正向/逆向云生成算法.通过将概念空间划分为几个子空间,然后再在各子空间上使用正向/逆向正态隶属云生成算法来拟合整个概念空间上问题的描述.在拟合过程中,子空间的划分数目是影响拟合效果的主要因素.通过一系列实验分析就子空间的划分数目给出了一个建议数目.最后,实例分析和比较结果表明了该算法的可行性和效率.     

基于数据域描述的模糊支持向量回归

针对支持向量机中由于噪声和孤立点带来的过拟合问题，提出了一种基于支持向量数据域描述的模糊隶属度函数模型，根据样本到特征空间最小包含超球球心的距离来确定其模糊隶属度．将提出的隶属度模型用于模糊支持向量回归中，二维数据集仿真以及工业PTA氧化过程中4-CBA浓度预测的实例表明，提出的模型可以有效减小回归误差，提高支持向量机抗噪声的能力．     

改进的KMSE方法及其实现

依据KMSE模型对应的特征空间中的鉴别矢量可表示为部分训练样本的线性组合这一理论前提,可利用回归分析中变量选择的思路对KMSE模型加以改进.在本文中为了提高KMSE的分类效率而发展出的基于最小平方误差准则的算法能大大提升KMSE模型的分类速度.实验结果显示该算法还能取得较优的分类性能.     

局部离群点挖掘算法研究

离群点可分为全局离群点和局部离群点.在很多情况下,局部离群点的挖掘比全局离群点的挖掘更有意义.现有的基于局部离群度的离群点挖掘算法存在检测精度依赖于用户给定的参数、计算复杂度高等局限.文中提出将对象属性分为固有属性和环境属性,用环境属性确定对象邻域、固有属性计算离群度的方法克服上述局限;并以空间数据为例,将空间属性与非空间属性分开,用空间属性确定空间邻域,用非空间属性计算空间离群度,设计了空间离群点挖掘算法.实验结果表明,所提算法具有对用户依赖性少、检测精度高、可伸缩性强和运算效率高的优点.     

基于相对密度的军事高维数据噪声点检测方法

军事训练领域的特殊性造成其相关数据存在大量的噪声点,同时也为噪声检测算法提出了相应的要求。分析现有数据噪声点检测算法,提出将数据属性分为空间属性、环境属性、特征属性,利用空间属性确定数据对象的分类,利用环境属性确定具有不同特征的数据对象邻域,利用特征属性计算离群度。改进了基于相对密度的离群度计算方法,提出LRDF算法,实验结果表明,该方法有效地提高了噪声点检测的精度和效率,增强算法可用性。     

基于边界区域局部模糊增强的uppiRKM 聚类算法

如何对交叉边界区域的数据对象进行度量与处理一直是粗糙k-means(RKM)及其衍生算法的主要出发点.uppiRKM算法通过引入Laplace无差别原则,较好地解决了传统RKM算法对权重系数的选择比较敏感等相关问题,但没有考虑边界区域多个类簇的交叉程度以及边界区域数据对象的空间位置分布对聚类结果的影响.鉴于此,设计一种对边界区域的数据对象进行局部模糊度量的方法,并提出基于边界区域局部模糊增强的uppiRKM聚类改进算法,通过多组实例分析验证了所提出算法的有效性.     

面向ROLAP的时态垂直粗分区方法

数据仓库为海量数据上的决策支持提供了一个高效的信息管理平台,ROLAP利用关系型数据仓库操纵灵活和技术成熟等优势,为面向数据仓库的分析和决策提供了有效的存取、建模和操作方法.然而,传统关系存取方法造成ROLAP的I/O有效性面临严峻的挑战.首先通过分析DSS应用的特点,提出了关系算子访问基表属性的时态行为,定义了算子对属性的时态局部访问.通过对查询样本集的解析建立算子与属性的时态访问映射矩阵,将有效增益作为属性的聚类准则得到时态访问模型PD.最后,给出了求解该模型的粗集算法以及依据聚类结果设计的基表属性的垂直分区方案.实验证明:在决策支持应用中,该方法的效率优于同类的其它优化分区方法.     

Clustering spatial data with a geographic constraint: exploring local search

Spatial data objects that possess attributes in the optimization domain and the geographic domain are now widely available. For example, sensor data are one kind of spatial data objects. The location of a sensor is an attribute in the geographic domain, while its reading is an attribute in the optimization domain. Previous studies discuss dual clustering problems that attempt to partition spatial data objects into several groups, such that objects in the same group have similar values in their optimization attributes and form a compact region in the geographic domain. However, previous studies do not clearly define compact regions. Therefore, this paper formulates a connective dual clustering problem with an explicit connected constraint given. Objects with a geographic distance smaller than or equal to the connected constraint are connected. The goal of the connective dual clustering problem is to derive clusters that contain objects with similar values in the optimization domain and are connected in the geographic domain. This study further proposes an algorithm CLS (Clustering with Local Search) to efficiently derive clusters. This algorithm consists of two phases: the ConGraph (standing for Connective Graph) transformation phase and the clustering phase. In the ConGraph transformation phase, CLS first transforms the data objects into a ConGraph that captures geographic constraints among data objects and selects initial seeds for clustering. Then, the initial seeds selected nearby data objects and formed coarse clusters by exploring local search in the clustering phase. Moreover, coarse clusters are merged and finely turned. Experiments show that CLS algorithm is more efficient and scalable than existing methods.     

Hilbert填充曲线与空间分布模式探测的点数据集空间划分方法

空间数据划分是空间索引、并行GIS数据分解以及分布式数据管理与调度等问题的核心环节之一。针对点数据集多目标空间划分问题,引入Hilbert空间填充曲线和空间分布模式探测过程,提出针对规则、随机和聚集分布模式的点数据集空间划分方法。实验结果表明,该方法能够在缺少覆盖范围信息的条件下准确判定空间分布类型,该方法能够兼顾空间聚集性、数据量均衡与空间重叠度3种约束条件。     

Spatial match retrieval based on direction signatures using multiple key hashing scheme

The spatial relationships among pictorial objects are important spatial characteristics in image database systems. Based on the concept of 9-DLT representation, we propose a new method for effective storage utilization and picture retrieval. First, we transform a symbolic picture into a set of triples associated with pairwise spatial relationships among objects. Then, an associated normalized signature record with nine attributes is specified. Finally, each normalized record of this image database can be well allocated by using our proposed multiple key hashing scheme and can be fast accessed for spatial match retrieval. The performance formula for spatial picture retrieval is presented and the theoretically optimal solutions to the bit lengths of all the attributes are derived in certain case. Also, we give a systematic method to solve the feasible bit lengths for the attributes based on the method proposed by Chang. © 1997 John Wiley & Sons, Inc.     

The data type of spatial objects

A spatial object consists of data assigned to points in a space. Spatial objects, such as memory states and three dimensional graphical scenes, are diverse and ubiquitous in computing. We develop a general theory of spatial objects by modelling abstract data types of spatial objects as topological algebras of functions. One useful algebra is that of continuous functions, with operations derived from operations on space and data, and equipped with the compact-open topology. Terms are used as abstract syntax for defining spatial objects and conditional equational specifications are used for reasoning. We pose a completeness problem: Given a selection of operations on spatial objects, do the terms approximate all the spatial objects to arbitrary accuracy? We give some general methods for solving the problem and consider their application to spatial objects with real number attributes.     

Knowledge Discovery in Multiple Spatial Databases

In this paper, a new approach for centralised and distributed learning from spatial heterogeneous databases is proposed. The centralised algorithm consists of a spatial clustering followed by local regression aimed at learning relationships between driving attributes and the target variable inside each region identified through clustering. For distributed learning, similar regions in multiple databases are first discovered by applying a spatial clustering algorithm independently on all sites, and then identifying corresponding clusters on participating sites. Local regression models are built on identified clusters and transferred among the sites for combining the models responsible for identified regions. Extensive experiments on spatial data sets with missing and irrelevant attributes, and with different levels of noise, resulted in a higher prediction accuracy of both centralised and distributed methods, as compared to using global models. In addition, experiments performed indicate that both methods are computationally more efficient than the global approach, due to the smaller data sets used for learning. Furthermore, the accuracy of the distributed method was comparable to the centralised approach, thus providing a viable alternative to moving all data to a central location.