正态分布和瑞利分布混合情形下的参数估计及分类问题

学者往往用单一的分布模拟和拟合杂波,如正态分布、瑞利分布和威布尔分布等。然而在实际中,雷达杂波由多种类型的杂波组成,单一分布通常不能精确刻画雷达杂波规律,因此,应用混合分布模型对雷达杂波数据建模更准确。本文考虑用正态分布和瑞利分布的混合分布拟合杂波,并应用矩估计方法和基于EM算法的极大似然估计方法估计模型参数,最后,应用最大后验概率分类准则验证2种估计方法的分类准确率。通过数据模拟,得出极大似然估计的效果和分类准确率都要优于矩估计的估计效果和分类准确率。     

基于自编码器的卷积神经网络肿瘤亚型分类研究

根据临床收录的肿瘤基因表达谱数据,可以利用分类器进行肿瘤亚型分类.由于基因表达谱数据样本小、维度高,难以提取有效特征,分类效果往往不好,而且很容易过拟合.针对这些问题,研究利用自编码器对特征基因进行降维,并构建多尺度的神经网络进行学习分类,综合考虑不同尺度的特征,提出A-CNNs网络,不仅解决了高维样本难以处理的问题,且有效避免了纵向加深神经网络带来的过拟合,得到了较高的平均分类精度,并与其他机器学习方法进行对比实验,实验证明所构建的分类模型可以取得较佳的分类效果.     

利用贝叶斯网络提高港口国检查效率

提高港口国监控(PSC)的检查效率,本文研究了船舶固有属性(船舶年龄、船旗、船级社、船舶尺度)、港口国检查缺陷项与船舶事故间的影响关系。本文所使用的数据主要来自于英国劳氏船级社(LR)、国际海事组织(IMO)和东京谅解备忘录(Tokyo MOU)三个数据库,共5478条干散货船数据。利用贝叶斯网络(BN)构建模型,并分别采用Bayesian Network (BN)和Greedy thick thinning(GTT)算法构建网络模型。同时利用K-折交叉验证、对数似然函数(LL)、赤池信息量准则(AIC)和贝叶斯信息准则(BIC)对模型进行评估。结果表明船舶的固有属性和关键检查缺陷项对船舶事故均有较高的直接影响,而大多数的港口国监控检查缺陷之间具有相互影响,并且通过关键检查缺陷项对船舶事故产生间接影响。因此可以利用关键检查缺陷项优化港口国检验制度,提高检验效率。     

负二项回归模型的推广及其在分类费率厘定中的应用

分类费率厘定中最常使用的模型之一是泊松回归模型,但当损失次数数据存在过离散特征时,通常会采用负二项回归模型。本文将两参数的负二项回归模型推广到了三参数情况,并用它来解决分类费率厘定中的过离散(over-dispersion)问题。本文通过对一组汽车保险损失数据的拟合表明,三参数的负二项分布回归模型可以有效改善对实际损失数据的拟合效果。     

纵向数据半参数回归模型的估计方法

本文考虑纵向数据半参数回归模型,通过考虑纵向数据的协方差结构,基于Profile最小二乘法和局部线性拟合的方法建立了模型中参数分量、回归函数和误差方差的估计量,来提高估计的有效性,在适当条件下给出了这些估计量的相合性.并通过模拟研究将该方法与最小二乘局部线性拟合估计方法进行了比较,表明了Profile最小二乘局部线性拟合方法在有限样本情况下具有良好的性质.     

基于机器学习的黏钢构件黏接层缺陷识别方法研究

对黏钢加固结构黏接层缺陷对超声检测信号的影响进行了深入研究,并提出了一种基于机器学习的黏接层缺陷识别的新型方法.首先,该文基于直接接触式的脉冲回波反射法对黏钢构件进行有限元模拟,并阐述了超声波在黏钢构件中的传播规律;其次,通过分析局部段超声回波信号及相关信号特征,讨论了不同缺陷变量对超声回波信号的影响规律;最后,建立了黏钢构件超声时程响应数据集,并对比了不同机器学习模型对缺陷大小、位置的分类识别性能,形成了黏钢构件黏接层缺陷识别方法.结果表明,局部段超声回波信号及其特征随着缺陷大小、位置的改变呈规律性变化,能够对缺陷信息进行初步区分.同时,该文提出的基于RF模型的黏钢构件黏接层缺陷识别方法能够有效识别黏钢构件黏接层缺陷,具有较广阔的工程应用前景.     

双向S-变精度优粗集模型及其应用

针对S-粗集模型及其现有扩展模型不能有效处理具有动态特性的偏好信息系统这一缺陷,本文提出了一种双向S-变精度优粗集模型,并给出了该模型的一些性质。新模型具有一定的抗数据干扰能力,从而可在动态偏好信息决策系统中获得更加合理的分类;最后给出一个应用实例,验证了新模型的有效性。     

基于凹凸组合的一类洛伦兹曲线模型

为了统计和分析一个国家和地区的收入分配情况,经济学界往往通过入户调查获得家庭收入与消费等数据,采用洛伦兹曲线模型来进行数据拟合.洛伦兹曲线模型拟合效果的好坏,直接影响着收入分配的描述.本文构建了一类凹凸组合的洛伦兹曲线模型,并针对19个国家的收入分配数据进行了实证分析.结果显示该模型具有较好的拟合效果,其基尼系数能较好地描述收入分配现状,对反映和监测居民之间的贫富差距具有重要意义.     

线性数据拟合方法的误差分析及其改进应用

利用最小二乘法进行线性数据拟合在一定条件下存在着误差较大的缺陷,为使线性数据拟合方法在科学实验和工程实践中能够更加准确地求解量与量之间的关系表达式,本文通过对常用线性数据拟合方法———最小二乘法进行了误差分析,并在此基础上提出了最小距离平方和法以对最小二乘法作改进处理.最后,通过举例分析对两种线性数据拟合方法的优劣加以讨论并分别给出其较为合理的应用控制条件.     

结构方程模型简化方法及在人机行为研究中的应用

为解决研究中结构方程模型过度复杂的问题,提出一种简化方法.分析了多维测量模型直接引入整体模型的局限性,对高阶验证性因子分析方法和项目组合法在模型简化中的适用性进行了分析,提出了首先检验简化必要性再使用高阶验证性因子分析方法和项目组合法对整体模型进行简化的两阶段方法.以电子学习系统持续使用行为的研究数据为例对该简化方法的有效性进行检验,简化前后拟合数据的对比分析表明该方法可以显著简化整体模型并获得理想的拟合优度.     

NHPP models for categorized software defects

We develop NHPP models to characterize categorized event data, with application to modelling the discovery process for categorized software defects. Conditioning on the total number of defects, multivariate models are proposed for modelling the defects by type. A latent vector autoregressive structure is used to characterize dependencies among the different types. We show how Bayesian inference can be achieved via MCMC procedures, with a posterior prediction‐based L‐measure used for model selection. The results are illustrated for defects of different types found during the System Test phase of a large operating system software development project. Copyright © 2005 John Wiley & Sons, Ltd.     

Calibrating software reliability models when the test environment does not match the user environment

Software failures have become the major factor that brings the system down or causes a degradation in the quality of service. For many applications, estimating the software failure rate from a user's perspective helps the development team evaluate the reliability of the software and determine the release time properly. Traditionally, software reliability growth models are applied to system test data with the hope of estimating the software failure rate in the field. Given the aggressive nature by which the software is exercised during system test, as well as unavoidable differences between the test environment and the field environment, the resulting estimate of the failure rate will not typically reflect the user‐perceived failure rate in the field. The goal of this work is to quantify the mismatch between the system test environment and the field environment. A calibration factor is proposed to map the failure rate estimated from the system test data to the failure rate that will be observed in the field. Non‐homogeneous Poisson process models are utilized to estimate the software failure rate in both the system test phase and the field. For projects that have only system test data, use of the calibration factor provides an estimate of the field failure rate that would otherwise be unavailable. For projects that have both system test data and previous field data, the calibration factor can be explicitly evaluated and used to estimate the field failure rate of future releases as their system test data becomes available. Copyright © 2002 John Wiley & Sons, Ltd.     

基于Kohonen网络的激光超声缺陷探测研究

在激光超声缺陷检测技术中,不同类型缺陷采样信号的准确分类至关重要.针对激光超声表面波实验采样信号高维小样本的特点,采用了一种有监督学习的Kohonen神经网络(S_Kohonen)自适应分类方法.在S_Kohonen网络自组织学习的过程中,通过改进网络的学习率提高了网络的收敛速度.通过采用一种无需邻域半径判断的自适应权值调整方式来实现竞争层神经元权值不同程度的调整,从而更有效的表征输入样本的分布特征.通过对不同类型缺陷探测样本的多次实验,验证了所述方法具有良好的分类预测效果,多次交叉验证分类正确率均能达到100%.     

基于软件测试的软件缺陷预测模型

通过对某武器系统软件测试所得缺陷数据与软件规模之间关系的分析,建立了基于软件测试的软件缺陷预测模型.通过本项研究,可以在武器系统软件研制的初期阶段对软件可能含有的缺陷数量和缺陷等级进行预测,为软件研制过程的质量保证活动策划提供依据.     

���������Ϣϵ�������ȱ��Ԥ��ģ��

In software defect prediction with a regression model, too many metrics extracted from static code and aggregated (sum, avg, max, min) from methods into classes can be candidate features, and the classical feature selection methods, such as AIC, BIC, should be processed at a given model. As a result, the selected feature sets are significantly different for various models without a reasonable interpretation. Maximal information coefficient (MIC) presented by Reshef et al.\ucite{4} is a novel method to measure the degree of the interdependence between two continuous variables, and an available computing method is also given based on the observations. This paper firstly use the MIC between defect counts and each feature to select features, and then conduct the power transformation on the selected features, and finally build up the principal component Poisson and negative binomial regression model. All experiments are conducted on KC1 data set in NASA repository on the level of class. The block-regularized $m\times 2$ cross-validated sequential $t$-test is employed to test the difference of performance of two models. The performance measures of a model in this paper are FPA, AAE, ARE. The experimental results show that 1) the aggregated features, such as sum, avg, max, are selected by MIC except min, which are significantly different from AIC, BIC; 2) the power transformation to the features can improve the performance for majority of models; 3) after PCA and factorial analysis, two clear factors are obtained in the model. One corresponds to the aggregated features via avg and max, and the other corresponds to the aggregated features with sum. Therefore, the model owns a reasonable interpretation. Conclusively, the aggregated features with sum, avg, max are significantly effective for software defect prediction, and the regression model based on the selected features by MIC has some advantages.     

Software reliability assessment models based on cumulative bernoulli trial processes

The binomial software reliability growth model (SRGM) contains most existing SRGMs proposed in earlier work as special cases, and can describe every software failure-occurrence pattern in continuous time. In this paper, we propose generalized binomial SRGMs in both continuous and discrete time, based on the idea of cumulative Bernoulli trials. It is shown that the proposed models give some new unusual discrete models as well as the well-known continuous SRGMs. Through numerical examples with actual software failure data, two estimation methods for model parameters with grouped data are provided, and the predictive model performance is examined quantitatively.     

Modelling the process of incoming problem reports on released software products

For big software developing companies, it is important to know the amount of problems of a new software product that are expected to be reported in a period after the date of release, on a weekly basis. For each of a number of past releases, weekly data are present on the number of such reports. Based on the type of data that is present, we construct a stochastic model for the weekly number of problems to be reported. The (non‐parametric) maximum likelihood estimator for the crucial model parameter, the intensity of an inhomogeneous Poisson process, is defined. Moreover, the expectation maximization algorithm is described, which can be used to compute this estimate. The method is illustrated using simulated data. Copyright © 2004 John Wiley & Sons, Ltd.     

Model of a spatially inhomogeneous one-dimensional active medium

We investigate the dynamics of one-dimensional discrete models of a one-component active medium analytically. The models represent spatially inhomogeneous diffusively concatenated systems of one-dimensional piecewise-continuous maps. The discontinuities (the defects) are interpreted as the differences in the parameters of the maps constituting the model. Two classes of defects are considered: spatially periodic defects and localized defects. The area of regular dynamics in the space of the parameters is estimated analytically. For the model with a periodic inhomogeneity, an exact analytic partition into domains with regular and with chaotic types of behavior is found. Numerical results are obtained for the model with a single defect. The possibility of the occurrence of each behavior type for the system as a whole is investigated. Translated from Teoreticheskaya i Matematicheskaya Fizika, Vol. 124, No. 3, pp. 506–519, September, 2000.     

A uniform framework for the combination of penalties in generalized structured models

Penalized estimation has become an established tool for regularization and model selection in regression models. A variety of penalties with specific features are available and effective algorithms for specific penalties have been proposed. But not much is available to fit models with a combination of different penalties. When modeling the rent data of Munich as in our application, various types of predictors call for a combination of a Ridge, a group Lasso and a Lasso-type penalty within one model. We propose to approximate penalties that are (semi-)norms of scalar linear transformations of the coefficient vector in generalized structured models—such that penalties of various kinds can be combined in one model. The approach is very general such that the Lasso, the fused Lasso, the Ridge, the smoothly clipped absolute deviation penalty, the elastic net and many more penalties are embedded. The computation is based on conventional penalized iteratively re-weighted least squares algorithms and hence, easy to implement. New penalties can be incorporated quickly. The approach is extended to penalties with vector based arguments. There are several possibilities to choose the penalty parameter(s). A software implementation is available. Some illustrative examples show promising results.     

Intelligent data analysis and model interpretation with spectral analysis fuzzy symbolic modeling

This paper proposes fuzzy symbolic modeling as a framework for intelligent data analysis and model interpretation in classification and regression problems. The fuzzy symbolic modeling approach is based on the eigenstructure analysis of the data similarity matrix to define the number of fuzzy rules in the model. Each fuzzy rule is associated with a symbol and is defined by a Gaussian membership function. The prototypes for the rules are computed by a clustering algorithm, and the model output parameters are computed as the solutions of a bounded quadratic optimization problem. In classification problems, the rules’ parameters are interpreted as the rules’ confidence. In regression problems, the rules’ parameters are used to derive rules’ confidences for classes that represent ranges of output variable values. The resulting model is evaluated based on a set of benchmark datasets for classification and regression problems. Nonparametric statistical tests were performed on the benchmark results, showing that the proposed approach produces compact fuzzy models with accuracy comparable to models produced by the standard modeling approaches. The resulting model is also exploited from the interpretability point of view, showing how the rule weights provide additional information to help in data and model understanding, such that it can be used as a decision support tool for the prediction of new data.