建立启发式评价函数h(n)的一般方法

本文给出了在启发式图搜索策略中建立有效的启发式评价函数h(n)的两个一般方法:相似问题逼近法和函数逼近法,并对这两种方法进行了讨论。     

一种基于进化泛函网络的建模与函数逼近方法

提出了一种进化泛函网络的建模与函数逼近方法,该方法把泛函网络建模过程转变为结构和泛函参数的优化搜索过程,利用遗传规划设计泛函网络神经元函数,对网络结构和参数共存且相互影响的复杂解空间进行全局最优搜索,实现泛函网络结构和参数的共同学习,并用混合基函数实现目标函数的逼近,改变了人们通常用同类型基函数来实现目标函数逼近的方式.数值仿真结果表明,提出的网络建模与逼近方法具有较高的逼近精度.     

基于全局和局部搜索的整型权值神经网络混合学习算法

提出基于全局搜索(差分进化算法)和局部搜索(模式搜索)的混合型神经网络学习算法(DEPS),并采用查找逼近法对sigmoid函数进行优化。实验部分采用曲线逼近和纱线图片分类两个实验,并与基本差分进化算法(ODE)和可再生动态差分进化算法(RDDE)在算法效率和性能进行对比、验证,说明算法的有效性。最后对整型和浮点型神经网络进行速度测试比较,说明整型权值神经网络在计算速度上远远快于浮点型权值神经网络。经算法训练后的神经网络更适合于结构精简、速度快的嵌入式系统。     

BP神经网络算法的改进及收敛性分析

研究BP神经网络的数学理论,详细分析几种流行的BP神经网络学习算法的优缺点.针对一般BP算法收敛速度慢,易陷入局部极小值的缺陷,受Fletcher-Reeves线性搜索方法的指引,提出基于改进共轭梯度法的BP算法.从理论方面对算法进行深入的分析,介绍算法的详细思路和具体过程.并将算法训练后的BP神经网络运用到函数逼近中去.仿真结果表明,这种改进方案确实能够改善算法在训练过程中的收敛特性,而且提高收敛速度,取得令人满意的逼近效果.     

累积正态分布函数的逼近函数综述

Fisher z变换是一个显式的初等函数,用来逼近累积正态分布函数（标准正态分布的累积分布函数）。介绍了累积正态分布函数逼近函数的评价标准,对有代表性的逼近函数表达式及相应的最大距离误差值进行归纳总结。     

一种基于最优分段函数逼近的过程神经网络训练算法

提出了一种基于最优分段函数逼近的过程神经网络学习算法。将网络时变输入信号和连接权在一定精度下表示为分段函数的拟合形式,根据最小均方误差准则,构建 PN N基于函数基展开的训练算法。选择低阶分段函数作为基函数,利用其良好的柔韧逼近和光滑可导性质,快速实现网络待定参数对函数样本的自适应学习。网络训练中,只需迭代调整分段函数的连接系数,可有效减少模型中的参数冗余、提高PNN对实际问题的建模能力。     

一种基于遗传算法的数字曲线多边形逼近方法

提出了一种基于遗传算法的数字曲线多边形改进逼近方法。该方法针对规则形状数字曲线的多边形逼近问题,以二进制向量序列表示的染色体作为每一个对应的逼近多边形候选解,将简化前后多边形质心偏移误差以及各被替换线段欧氏距离的方差引入到适应函数中,用迭代次数的sigmoid函数作为变异概率来控制遗传算法优化求解过程中的全局和局部搜索特性。实验结果表明,该方法对于保持曲线多边形简化逼近后的形状特征具有较好的效果。     

基于狼群算法的短时高速交通流量预测仿真

由于传统的短时高速交通流量预测算法的受到函数逼近能力影响,极易陷入局部极值的问题,提出基于狼群算法的短时高速交通流量预测算法。构造型前馈小波神经网络算法在多维空间内对短时高速交通流量预测,需通过构造型前馈小波神经网络算法分析随机相关多维样本逼近能力,利用改进狼群算法的全局寻优能力,提升算法搜索精度,并完成目标的搜索,根据搜索结果获取优质的函数逼近值最优解,提升构造型前馈小波神经网络算法函数的逼近能力,获取最佳短时高速交通流量预测值。仿真结果显示,上述算法的短时高速交通流量预测效果显著优于小波神经网络预测算法,且本文算法的预测精度较高,具有较高的稳定性。     

多输入Sigmoid激励函数神经网络权值与结构确定法

结合伪逆直接计算得到神经元之间最优权值的方法,提出了一种双阶段自动搜索与确定最优网络结构的算法,克服了原有BP神经网络模型及其学习算法的固有缺陷。以函数逼近为例,计算机数值实验结果显示了算法有效且耗时短,证实了由该算法得到的网络对于多输入函数具有较优良的逼近(学习与校验)性能。     

基于熵误差函数Pi-Sigma神经网络在线梯度学习算法

Pi-Sigma神经网络因具有较强的非线性处理能力和较高的学习效率而受到广泛的关注。目前，该网络已经应用于函数逼近、信号处理、模式识别等领域。因此，提出了一种基于熵误差函数的Pi-Sigma神经网络在线学习算法，通过函数逼近及多维数据集分类实验验证了该算法的有效性。结果表明，所提算法在函数逼近、训练精度、测试精度和效率方面均有较好的表现。     

A parameter optimization approach to robust PI-controller design for systems with interval plants

The purpose of this paper is to develop an analytic way for designing optimal PI-controllers for the interval plant family. Optimality means that the coefficient intervals of the plant stabilized by a PI-controller is maximized. It will be shown that the optimization problem can be formulated in terms of an eigenvalue minimization problem of matrix pairs of the form （H（h0, g0）, H（δ1,k, δ2,k））, where k= 1, 2, 3, 4 and both H（h0, g0） and H（δ1,k,δ2,k） are frequency independent Hurwitz-like matrices. A numerical example is provided to illustrate that optimal controller parameters can be successfully obtained by a two-dimensional search.     

Time-varying continuous nonlinear systems: uniform asymptotic stability

The framework of the presented research is a large class of time-varying nonlinear systems with continuous motions. The study of the uniform asymptotic stability of the zero equilibrium state developed here goes back to, and relies on, the very foundations of the Lyapunov stability concept and the (second) Lyapunov method. Stability domains are first defined and examined. Then, their qualitative features are used to establish complete solutions to the problem of uniform asymptotic stability of the equilibrium for various subclasses of the systems. The solutions present conditions that are both necessary and sufficient for: (1) the uniform asymptotic stability, (2) an exact direct one-shot construction of a system Lyapunov function and (3) for a direct accurate one-shot determination of the asymptotic stability domain. In addition, the solutions establish a novel Lyapunov-method based approach to the asymptotic stability analysis. This enables an arbitrary selection of a function p(·) from a defined functional family to determine a Lyapunov function (·), [(·)], by solving D ⁺(·) = ? p(·) or, equivalently, D ⁺ (·) = ? p(·)[1 ? (·)], respectively. Illustrative examples are presented.     

A simple,accurate method to compute brightness temperature and/or radiance for the thermal infrared channels of the Advanced Very High Resolution Radiometer (AVHRR)

Users of thermal infrared data from the AVHRR on a NOAA polar-orbiting operational satellite convert the count value output to radiance units, and then assign an equivalent blackbody temperature to the radiance value. Assigning a blackbody temperature to the radiance value is an indirect process, which requires knowledge of the AVHRR spectral response function and a fairly complex calculation. Both difficulties can be avoided by the simple two-step process shown in this Letter. First, blackbody temperature is estimated from a square-root of the measured radiance, then the estimate is refined by values from a ‘universal’ correction curve. The RMS difference between this approximation and the complex calculation is a few hundredths deg K for temperatures in the 200-320 deg K range. The inverse computation, radiance from temperature, is accurate to within 0·01-0·02mWm^?2sr^?1 (cm^?1)^?1. Results are shown for the NOAA-7, -9, -11, and -12 spacecraft.     

Estimating soil moisture from 6·6 GHz dual polarization,and/or satellite derived vegetation index

Eight and a half years (January 1979 to August 1987) of Scanning Multichannel Microwave Radiometer (SMMR) data taken at a frequency of 6·6 GHz for both day and night observations at both polarizations were processed, documented and used to study the relationship between brightness temperature (T_B) and antecedent precipitation index (API) in a wide range of vegetation index (normalized difference vegetation index (NDVI) varies from 0·2 to 0·6) in the mid-west and southern United States. In general, this study validates the model structure for soil wetness developed by Choudhury and Golus. For NDVI greater than 0·45 the resultant microwave signal is substantially affected by the vegetation. The night-time observations by both polarizations gave a better correlation between T_B and API. The horizontal polarization is more sensitive to vegetation. For the least and greatest vegetated areas, nighttime observations by vertical polarization showed less scatter in the T_B versus API relation. A non-linear model was developed for soil wetness using horizontal and vertical plarization and their difference. The estimate of error for this model is better than previous models, and can be used to obtain six levels of soil moisture.     

A measurement fusion method for nonlinear system identification using a cooperative learning algorithm

Identification of a general nonlinear noisy system viewed as an estimation of a predictor function is studied in this article. A measurement fusion method for the predictor function estimate is proposed. In the proposed scheme, observed data are first fused by using an optimal fusion technique, and then the optimal fused data are incorporated in a nonlinear function estimator based on a robust least squares support vector machine (LS-SVM). A cooperative learning algorithm is proposed to implement the proposed measurement fusion method. Compared with related identification methods, the proposed method can minimize both the approximation error and the noise error. The performance analysis shows that the proposed optimal measurement fusion function estimate has a smaller mean square error than the LS-SVM function estimate. Moreover, the proposed cooperative learning algorithm can converge globally to the optimal measurement fusion function estimate. Finally, the proposed measurement fusion method is applied to ARMA signal and spatial temporal signal modeling. Experimental results show that the proposed measurement fusion method can provide a more accurate model.     

Biomechanical comparison of carpet-stretching devices

Peak impact forces, measured from seven male carpetlayers using a knee-kicker over 39 trials were 2933?N (SD 397), taking 9·5?ms (SD ± 1·1) to reach peak. The knee-kicking cycle involved high knee decelerations of 880?m.s^?2 (SD ±271). Angles at peak force for the knee (63° SD ± 10°) and hip (80° SD ± 35°) showed considerable variation between individuals. Kneeling forces on the non-kicking leg during the kick cycle were 894?N: comparative forces were 368?N and 476?N measured during use of the power stretcher and crab re-stretcher, respectively. Under dynamic conditions the effective mechanical advantage of the knee-kicker was less than 1·0. In comparison, the mechanical advantage of the power stretcher was approximately 14 times greater over the stretch action. When tested with a crab re-stretcher, the breaking strength of the smooth-edge to which the carpet attaches averaged 2384?N (SD ± 245). Design guidelines for an improved carpet stretching device are provided.     

The value of software sizing

Context

One of the difficulties faced by software development Project Managers is estimating the cost and schedule for new projects. Previous industry surveys have concluded that software size and cost estimation is a significant technical area of concern. In order to estimate cost and schedule it is important to have a good understanding of the size of the software product to be developed. There are a number of techniques used to derive software size, with function points being amongst the most documented.

Objective

In this paper we explore the utility of function point software sizing techniques when applied to two levels of software requirements documentation in a commercial software development organisation. The goal of the research is to appraise the value (cost/benefit) which functional sizing techniques can bring to the project planning and management of software projects within a small-to-medium sized software development enterprise (SME).

Method

Functional counts were made at the bid and detailed functional specification stages for each of five commercial projects used in the research. Three variants of the NESMA method were used to determine these function counts. Through a structured interview session, feedback on the sizing results was obtained to evaluate its feasibility and potential future contribution to the company.

Results

The results of our research suggest there is value in performing size estimates at two appropriate stages in the software development lifecycle, with simplified methods providing the optimal return on effort expended.

Conclusion

The ‘Estimated NESMA’ is the most appropriate tool for use in size estimation for the company studied. The use of software sizing provides a valuable contribution which would augment, but not replace, the company’s existing cost estimation approach.     

On a Theoretical Justification of the Choice of Epsilon-Inflation in PASCAL-XSC

In many interval computation methods, if we cannot guarantee a solution within a given interval, it often makes sense to "inflate" this interval a little bit. There exist many different "inflation" methods. The authors of PASCAL-XSC, after empirically comparing the behavior of different inflation methods, decided to implement the formula [x-,x+]ε = [(1 + ε)x- - ε · x+, (1 + ε)x+ - ε · x-]. A natural question is: Is this choice really optimal (in some reasonable sense), or is it only an empirical approximation to the truly optimal choice? In this paper, we show that this empirical choice can be theoretically justified. Namely, we will give two justifications: First, the inflation method used in PASCAL-XSC is the only inflation that is invariant w.r.t. some reasonable symmetries; and Second, that this inflation method is optimal in some reasonable sense.     

Randomized Algorithms for 3-SAT

Schoning proposed a simple yet efficient randomized algorithm for solving the k-SAT problem. In the case of 3-SAT, the algorithm has an expected running time of poly(n)·(4/3)ⁿ = O(1.334ⁿ). In this paper we present randomized algorithms and show that one of them has O(1.3302ⁿ) expected running time, improving Schoning's algorithm. (Note. At this point, the fastest randomized algorithm for 3-SAT is the one given by Iwama and Tamaki that runs in O(1.324ⁿ).)     

Reliability analysis and optimal version-updating for open source software

Context

Although reliability is a major concern of most open source projects, research on this problem has attracted attention only recently. In addition, the optimal version-dating for open source software considering its special properties is not yet discussed.

Objective

In this paper, the reliability analysis and optimal version-updating for open source software are studied.

Method

A modified non-homogeneous Poisson process model is developed for open source software reliability modeling and analysis. Based on this model, optimal version-updating for open source software is investigated as well. In the decision process, the rapid release strategy and the level of reliability are the two most important factors. However, they are essentially contradicting with each other. In order to consider these two conflicting factors simultaneously, a new decision model based on multi-attribute utility theory is proposed.

Results

Our models are tested on the real world data sets from two famous open source projects: Apache and GNOME. It is found that traditional software reliability models provide overestimations of the reliability of open source software. In addition, the proposed decision model can help management to make a rational decision on the optimal version-updating for open source software.

Conclusion

Empirical results reveal that the proposed model for open source software reliability can describe the failure process more accurately. Furthermore, it can be seen that the proposed decision model can assist management to appropriately determine the optimal version-update time for open source software.