聚类优化RBF网络在锅炉燃烧中的应用

为了提高径向神经网络的训练精度,提出一种混合优化算法.算法将基于萤火虫算法的模糊聚类,应用到径向神经网络基函数中心向量的计算中,利用萤火虫算法良好的全局寻优能力来优化搜索基函数中心,提高了获取网络类中心的稳定性.锅炉燃烧优化的实例表明,混合优化算法达到了预期效果,提升了锅炉燃烧效率.     

模糊神经网络在多源数据融合中的应用研究

在T-S模糊神经网络数据融合的基础上,改进了标准T-S模糊融合算法中的模糊算子,并利用聚类算法对网络结构中模糊隶属度个数进行选取.通过仿真实验,验证了改进的算法在融合过程中的合理性、稳定性和准确性.以及聚类算法在T-S模糊神经网络数据融合算法中运用的合理性和有效性.     

模糊C均值算法的改进

模糊聚类分析方法具有较强的实用性,但传统的模糊C均值算法对数据集进行分类时有均分的趋势,对于数据集中各类样本数目相差较大的情况,其聚类结果不是很理想.因此,本文对FCM算法进行了改进,使之不但能够达到更好的分类效果,同时也更加适用于样本分类不均衡的聚类问题.文中还结合具体算例进行了聚类分析,得到了理想的分类效果.     

基于模糊聚类分析的小波神经网络模型在径流预测中的应用

提出了一种在对预报因子集进行模糊聚类分析基础上构建径流预测模型的新方法:先通过模糊C-均值聚类将历史径流数据进行分类,然后利用小波神经网络分别建立预报因子集类别变量特征值与观测值之间的局部预测模型,并设计了特征值分类识别器,自动搜寻相适应的局部网络模型进行预测.通过西南某水库2011年日平均入库来流的计算实例对简单小波神经网络预测模型和所建的基于FCM与小波神经网络的预测模型进行了比较,结果较为满意.     

FCM聚类算法中模糊加权指数m的优选方法

模糊c-均值(FCM)聚类算法是一种通过目标函数的极小化来获得数据集模糊划分的方法。其中,模糊加权指数m对FCM算法的分类性能有着重要的影响,而调用FCM算法进行模糊聚类分析时又必须给m赋值。因此,模糊加权指数m的优选研究就变得很有意义。基于模糊决策的方法本文给出了一种对m的优选方法,实验结果表明该方法是有效的。     

土壤分类的模糊集成算法与应用

土壤是一个多性状的连续体,其分类的首选方法是模糊聚类分析.但是模糊聚类分析中现有的基于模糊等价关系的动态聚类法和模糊c-均值法各有利弊,采用其中一种方法聚类肯定存在不足.为此集成两种聚类方法的优点,避其缺点,提出了用基于模糊等价关系的动态聚类方法和方差分析方法确定聚类数目和初始聚类中心,再用模糊c-均值法决定最终分类结果的集成算法,并将其应用到松花江流域土壤分类中,得到了较为切合实际的分类结果.     

模糊ISODATA中分类数C的确定

模糊ISODATA方法作为一种常用的聚类方法,在许多领域得到了广泛的应用.但作为一种聚类方法,它却没有显示出能自动聚类的效果,而只是对普通分类的一个扩展.在每次聚类前它仍要预先凭经验来给定一个分类数C因此,本文针对这个缺陷,提出了一个改进方案,使之能自动给出分类数C并在实例中得到了检验和应用.     

基于随机采样模糊聚类和卡尔曼滤波方法的模糊辨识

提出一种新的基于模糊聚类和卡尔曼滤波方法的模糊辨识算法 .该方法是基于快速模糊聚类 ,计算给定样本在各类中的隶属度 ,并利用卡尔曼滤波方法辨识模糊模型的结论参数 .整个辨识过程与一般的模糊聚类方法 [1 ]相比 ,需要的 CPU时间大大缩短 .最后通过仿真实例验证了该方法的有效性 .     

基于微分进化算法的FCM图像分割算法

为提高模糊C均值(FCM)算法的自动化程度,提出基于微分进化算法的FCM图像分割算法(DEFCM),利用微分进化算法全局性和鲁棒性的特点自动确定分类数和初始聚类中心,再将其作为模糊c均值聚类的初始聚类中心,弥补FCM算法的不足.实验表明该算法不仅能够正确地对图像分类,而且能获得较好的图像分割效果和质量.     

模糊数空间上的积分度量及其在模糊聚类中的应用

在本文中,首先利用区间数的EW-型度量探讨了模糊数空间上的积分度量问题,给出了模糊数空间上的一种新的积分度量-EW-型积分度量,并证明了其相关性质.其次,作为EW-型积分度量的应用,设计了对属性特征为三角模糊数的事物进行分类的模糊聚类算法.然后通过实例分析,说明了EW-型积分度量使模糊聚类算法实现的更简单易行,分类更加精细,合理有效等。     

基于DDAG-SVM的网络流量分类技术

互联网技术不断发展,很多新的网络通信采用动态端口、协议加密等技术,使传统的流量分类技术不再适用.以TCP三次握手后客户端到服务器的第1个包载荷大小、服务器到客户端的第1个包和第2个包载荷大小以及服务器端口信息作为流量特征,提出一种基于DDAG-SVM的网络流量分类的方法,并针对传统DDAG-SVM的误差累积效应,使分类性能变差的问题,根据类间可分离度重构DDAG-SVM决策树,每次都选择最容易分开的两个流类别构成分类决策面,测试结果表明该方法取得了较高的分类准确率.     

Direct versus indirect credit scoring classifications

We introduce a new approach to assigning bank account holders to ‘good’ or ‘bad’ classes based on their future behaviour. Traditional methods simply treat the classes as qualitatively distinct, and seek to predict them directly, using statistical techniques such as logistic regression or discriminant analysis based on application data or observations of previous behaviour. We note, however, that the ‘good’ and ‘bad’ classes are defined in terms of variables such as the amount overdrawn at the time at which the classification is required. This permits an alternative, ‘indirect’, form of classification model in which, first, the variables defining the classes are predicted, for example using regression, and then the class membership is derived deterministically from these predicted values. We compare traditional direct methods with these new indirect methods using both real bank data and simulated data. The new methods appear to perform very similarly to the traditional methods, and we discuss why this might be. Finally, we note that the indirect methods also have certain other advantages over the traditional direct methods.     

Admissible Transformations and Normalized Classes of Nonlinear Schrödinger Equations

The theory of group classification of differential equations is analyzed, substantially extended and enhanced based on the new notions of conditional equivalence group and normalized class of differential equations. Effective new techniques are proposed. Using these, we exhaustively describe admissible point transformations in classes of nonlinear (1+1)-dimensional Schrödinger equations, in particular, in the class of nonlinear (1+1)-dimensional Schrödinger equations with modular nonlinearities and potentials and some subclasses thereof. We then carry out a complete group classification in this class, representing it as a union of disjoint normalized subclasses and applying a combination of algebraic and compatibility methods. Moreover, we introduce the complete classification of (1+2)-dimensional cubic Schrödinger equations with potentials. The proposed approach can be applied to studying symmetry properties of a wide range of differential equations.     

Novel design model for the stator currents subsystem of induction motors

The stator currents subsystem is a vital element of many high-performance induction motor control schemes. While there are several control techniques available for this subsystem, traditional linear controllers are still widely used because of its simplicity and proven effectiveness. However, the traditional simplified design-model lacks important information, necessary for the design of high-performance and robust controllers. In this article a novel design-model intended for linear controller formulation and evaluation is developed. This new mathematical representation captures several elements which are missing in the traditional representation, maintaining at the same time a similar level of simplicity. Along the derivation of this new representation several models of decreasing complexity and comprehensiveness are also presented together with a critical classification. This classification is intended to aid the designer in choosing the appropriate mathematical representation for specific purposes. Finally, the article is accompanied with experimental findings which illustrate the use of the proposed model.     

Two‐group classification using the Bayesian data reduction algorithm

The Bayesian data reduction algorithm (BDRA) is compared to traditional classification methods as well as feed forward artificial neural networks through a rigorous experiment. The BDRA performs comparably to alternative techniques and approaches theoretical optimal classification rates. Furthermore, it has a fundamentally different method for determining class membership. This study is novel in that it explores how the BDRA relates to established techniques, how it might be used in an explanatory manner, and how best to use it. © 2009 Wiley Periodicals, Inc. Complexity, 2010     

Discrimination with jointly equicorrelated multi-level multivariate data

In this article we study a linear as well as a quadratic discriminant function for multi-level multivariate repeated measurement data under the assumption of multivariate normality. We assume that the m-variate observations have jointly equicorrelated covariance structure in addition to a Kronecker product structure on the mean vector. The new discriminant functions are very effective in discriminating individuals when the number of observations is very small. The proposed classification rules are demonstrated on a real data set. The error rates of the proposed classification rules are found to be much less than the error rates of the traditional classification rules, when in fact the traditional classification rules fail most of the time owing to the small sample sizes.     

On the socles of fully invariant subgroups of Abelian p-groups

The classification of the fully invariant subgroups of a reduced Abelian p-group is a difficult long-standing problem when one moves outside of the class of fully transitive groups. In this work we restrict attention to the socles of fully invariant subgroups and introduce a new class of groups which we term socle-regular groups; this class is shown to be large and strictly contains the class of fully transitive groups. The basic properties of such groups are investigated but it is shown that the classification of even this simplified class of groups, seems extremely difficult. Received: 4 September 2008     

Multiclass Probability Estimation With Support Vector Machines

Multiclass classification and probability estimation have important applications in data analytics. Support vector machines (SVMs) have shown great success in various real-world problems due to their high classification accuracy. However, one main limitation of standard SVMs is that they do not provide class probability estimates, and thus fail to offer uncertainty measure about class prediction. In this article, we propose a simple yet effective framework to endow kernel SVMs with the feature of multiclass probability estimation. The new probability estimator does not rely on any parametric assumption on the data distribution, therefore, it is flexible and robust. Theoretically, we show that the proposed estimator is asymptotically consistent. Computationally, the new procedure can be conveniently implemented using standard SVM softwares. Our extensive numerical studies demonstrate competitive performance of the new estimator when compared with existing methods such as multiple logistic regression, linear discrimination analysis, tree-based methods, and random forest, under various classification settings. Supplementary materials for this article are available online.     

A New Invariant of Quadratic Lie Algebras

We define a new invariant of quadratic Lie algebras and give a complete study and classification of singular quadratic Lie algebras, i.e. those for which the invariant does not vanish. The classification is related to O(n)-adjoint orbits in $\mathfrak{o}(n)$ .     

Projection Pursuit for Exploratory Supervised Classification

In high-dimensional data, one often seeks a few interesting low-dimensional projections that reveal important features of the data. Projection pursuit is a procedure for searching high-dimensional data for interesting low-dimensional projections via the optimization of a criterion function called the projection pursuit index. Very few projection pursuit indices incorporate class or group information in the calculation. Hence, they cannot be adequately applied in supervised classification problems to provide low-dimensional projections revealing class differences in the data. This article introduces new indices derived from linear discriminant analysis that can be used for exploratory supervised classification.