Investigating ARIMA models of software system quality

In this paper, we investigate how to incorporate program complexity measures with a software quality model. We collect software complexity metrics and fault counts from each build during the testing phase of a large commercial software system. Though the data are limited in quantity, we are able to predict the number of faults in the next build. The technique we used is called times series analysis and forecasting. The methodology assumes that future predictions are based on the history of past observations. We will show that the combined complexity quality model is an improvement over the simpler quality only model. Finally, we explore how the testing process used in this development may be improved by using these predictions and suggest areas for future research.     

An empirical study of program quality during testing and maintenance

In this paper, we report the results of a study conducted on a large commercial software system written in assembly language. Unlike studies of the past, our data represent the unit test, integration, and all categories of the maintenance phase: adaptive, perfective, and corrective. The results confirm that faults and change activity are related to software measurements. In addition, we report the relationship between the number of design change requests and software measurements. This new observation has the potential to aid the software engineering management process. Finally, we demonstrate the value of multiple regression models over simple regression models.     

广义回归神经网络在软件质量预测中的应用

软件质量预测技术是软件质量评价体系中的关键技术,它能够对用户所关心的软件质量特性进行评价。广义回归神经网络在逼近能力、分类能力和学习速度方面具有较强优势。采用基于软件度量的广义回归神经网络构造质量预测模型能够从历史数据中寻找软件度量之间的相关关系。对软件缺陷数进行预测的实验说明了模型的有效性、精确性,实验结果令人满意。     

An investigation of artificial neural networks based prediction systems in software project management

A critical issue in software project management is the accurate estimation of size, effort, resources, cost, and time spent in the development process. Underestimates may lead to time pressures that may compromise full functional development and the software testing process. Likewise, overestimates can result in noncompetitive budgets. In this paper, artificial neural network and stepwise regression based predictive models are investigated, aiming at offering alternative methods for those who do not believe in estimation models. The results presented in this paper compare the performance of both methods and indicate that these techniques are competitive with the APF, SLIM, and COCOMO methods.     

Fault Prediction Modeling for Software Quality Estimation: Comparing Commonly Used Techniques

High-assurance and complex mission-critical software systems are heavily dependent on reliability of their underlying software applications. An early software fault prediction is a proven technique in achieving high software reliability. Prediction models based on software metrics can predict number of faults in software modules. Timely predictions of such models can be used to direct cost-effective quality enhancement efforts to modules that are likely to have a high number of faults. We evaluate the predictive performance of six commonly used fault prediction techniques: CART-LS (least squares), CART-LAD (least absolute deviation), S-PLUS, multiple linear regression, artificial neural networks, and case-based reasoning. The case study consists of software metrics collected over four releases of a very large telecommunications system. Performance metrics, average absolute and average relative errors, are utilized to gauge the accuracy of different prediction models. Models were built using both, original software metrics (RAW) and their principle components (PCA). Two-way ANOVA randomized-complete block design models with two blocking variables are designed with average absolute and average relative errors as response variables. System release and the model type (RAW or PCA) form the blocking variables and the prediction technique is treated as a factor. Using multiple-pairwise comparisons, the performance order of prediction models is determined. We observe that for both average absolute and average relative errors, the CART-LAD model performs the best while the S-PLUS model is ranked sixth.     

A fast procedure for the training of neural networks

A new training paradigm for artificial neural networks is described. The technique utilizes a polynomial approximation to the sigmoidal processing function and directly integrates principal components analysis (PCA) into the network training philosophy. A major benefit of the new technique is that off-line network training is ‘one-shot’, contrary to the standard iterative techniques available in the literature. Further training may be performed on-line in a recursive fashion, yielding an adaptive neural network. Additionally, the new philosophy incorporates a systematic procedure for determining the number of neurons in the hidden layer of the network. The training procedure is first described and the implications of the training philosophy discussed. Some results, including applications to industrial chemical processes, are then presented to highlight the power of the technique. The systems considered are a continuous stirred tank reactor and a polymerization reactor.     

Computational neural networks for mapping calorimetric data: Application of feed-forward neural networks to kinetic parameters determination and signals filtering

Feedforward neural networks have been used for kinetic parameters determination and signal filtering in differential scanning calorimetry. The proper learning function was chosen and the network topology was optimised using an empiric procedure. The learning process was achieved using various simulated thermoanalytical curves computed for several thermodynamic and kinetic parameters. Various amounts of simulated noise were added on the power signals. The resilient-propagation algorithm led to the best minimisation of the error computed over all the patterns. Relative errors on the thermodynamic and kinetic parameters were evaluated and compared to those obtained with the usual thermal analysis methods. The results obtained are very promising, and the errors are much lower than with usual methods, especially in the presence of noisy signals. This study shows that simulated thermoanalytical curves produced by Joule effect may be used for the deconvolution of the response of the apparatus, by using artificial neural networks.     

人工神经网络在预测服装企业安全库存的应用

安全库存是一种额外持有的库存，它作为一种缓冲器用来补偿在订货提前期内实际需求超过期望需求量或实际提前期超过期望提前期所产生的需求。在服装企业中一般凭经验来设定安全库存，但实际效果不佳，应用人工神经网络，建立BP神经网络模型，用多个影响安全库存的指标及安全库存对网络进行训练，以达到对安全库存量预测的目的，经验证和预测效果十分理想。     

Digitizing uncertainty modeling for reverse engineering applications: regression versus neural networks

The coordinate measuring machine is one of the two types of digitizers most popularly used in reverse engineering. A number of factors affect the digitizing uncertainty, such as travel speeds of the probe, pitch values (sampling points), probe angles (part orientations), probe sizes, and feature sizes. A proper selection of these parameters in a digitization or automatic inspection process can improve the digitizing accuracy for a given coordinate-measuring machine. To do so, some empirical models or decision rules are required. This paper applies and compares the nonlinear regression analysis and neural network modeling methods in developing empirical models for estimating the digitizing uncertainty. The models developed in this research can aid error prediction, accuracy improvement, and operation parameter selection in computer-aided reverse engineering and automatic inspection.     

The essential order of approximation for neural networks

There have been various studies on approximation ability of feedforward neural networks (FNNs). Most of the existing studies are, however, only concerned with density or upper bound estimation on how a multivariate function can be approximated by an FNN, and consequently, the essential approximation ability of an FNN cannot be revealed. In this paper, by establishing both upper and lower bound estimations on approximation order, the essential approximation ability (namely, the essential approximation order) of a class of FNNs is clarified in terms of the modulus of smoothness of functions to be approximated. The involved FNNs can not only approximate any continuous or integrable functions defined on a compact set arbitrarily well, but also provide an explicit lower bound on the number of hidden units required. By making use of multivariate approximation tools, it is shown that when the functions to be approximated are Lipschitzian with order up to 2, the approximation speed of the FNNs is uniquely deter     

Ranking importance of input parameters of neural networks

Artificial neural networks have been used for simulation, modeling, and control purposes in many engineering applications as an alternative to conventional expert systems. Although neural networks usually do not reach the level of performance exhibited by expert systems, they do enjoy a tremendous advantage of very low construction costs. This paper addresses the issue of identifying important input parameters in building a multilayer, backpropagation network for a typical class of engineering problems. These problems are characterized by having a large number of input variables of varying degrees of importance; and identifying the important variables is a common issue since elimination of the unimportant inputs leads to a simplification of the problem and often a more accurate modeling or solution. We compare three different methods for ranking input importance: sensitivity analysis, fuzzy curves, and change of MSE (mean square error); and analyze their effectiveness. Simulation results based on experiments with simple mathematical functions as well as a real engineering problem are reported. Based on the analysis and our experience in building neural networks, we also propose a general methodology for building backpropagation networks for typical engineering applications.     

Resource oriented selection of rule-based classification models: An empirical case study

The amount of resources allocated for software quality improvements is often not enough to achieve the desired software quality. Software quality classification models that yield a risk-based quality estimation of program modules, such as fault-prone (fp) and not fault-prone (nfp), are useful as software quality assurance techniques. Their usefulness is largely dependent on whether enough resources are available for inspecting the fp modules. Since a given development project has its own budget and time limitations, a resource-based software quality improvement seems more appropriate for achieving its quality goals. A classification model should provide quality improvement guidance so as to maximize resource-utilization. We present a procedure for building software quality classification models from the limited resources perspective. The essence of the procedure is the use of our recently proposed Modified Expected Cost of Misclassification (MECM) measure for developing resource-oriented software quality classification models. The measure penalizes a model, in terms of costs of misclassifications, if the model predicts more number of fp modules than the number that can be inspected with the allotted resources. Our analysis is presented in the context of our Rule-Based Classification Modeling (RBCM) technique. An empirical case study of a large-scale software system demonstrates the promising results of using the MECM measure to select an appropriate resource-based rule-based classification model. Taghi M. Khoshgoftaar is a professor of the Department of Computer Science and Engineering, Florida Atlantic University and the Director of the graduate programs and research. His research interests are in software engineering, software metrics, software reliability and quality engineering, computational intelligence applications, computer security, computer performance evaluation, data mining, machine learning, statistical modeling, and intelligent data analysis. He has published more than 300 refereed papers in these areas. He is a member of the IEEE, IEEE Computer Society, and IEEE Reliability Society. He was the general chair of the IEEE International Conference on Tools with Artificial Intelligence 2005. Naeem Seliya is an Assistant Professor of Computer and Information Science at the University of Michigan - Dearborn. He recieved his Ph.D. in Computer Engineering from Florida Atlantic University, Boca Raton, FL, USA in 2005. His research interests include software engineering, data mining and machine learnring, application and data security, bioinformatics and computational intelligence. He is a member of IEEE and ACM.     

Lower estimation of approximation rate for neural networks

Let SFd and Πψ,n,d = { nj=1bjψ(ωj·x+θj) :bj,θj∈R,ωj∈Rd} be the set of periodic and Lebesgue’s square-integrable functions and the set of feedforward neural network (FNN) functions, respectively. Denote by dist (SF d, Πψ,n,d) the deviation of the set SF d from the set Πψ,n,d. A main purpose of this paper is to estimate the deviation. In particular, based on the Fourier transforms and the theory of approximation, a lower estimation for dist (SFd, Πψ,n,d) is proved. That is, dist(SF d, Πψ,n,d) (nlogC2n)1/2 . T...     

Performance of radial basis and LM-feed forward artificial neural networks for predicting daily watershed runoff

This study investigated the effects of upstream stations’ flow records on the performance of artificial neural network (ANN) models for predicting daily watershed runoff. As a comparison, a multiple linear regression (MLR) analysis was also examined using various statistical indices. Five streamflow measuring stations on the Cahaba River, Alabama, were selected as case studies. Two different ANN models, multi layer feed forward neural network using Levenberg–Marquardt learning algorithm (LMFF) and radial basis function (RBF), were introduced in this paper. These models were then used to forecast one day ahead streamflows. The correlation analysis was applied for determining the architecture of each ANN model in terms of input variables. Several statistical criteria (RMSE, MAE and coefficient of correlation) were used to check the model accuracy in comparison with the observed data by means of K-fold cross validation method. Additionally, residual analysis was applied for the model results. The comparison results revealed that using upstream records could significantly increase the accuracy of ANN and MLR models in predicting daily stream flows (by around 30%). The comparison of the prediction accuracy of both ANN models (LMFF and RBF) and linear regression method indicated that the ANN approaches were more accurate than the MLR in predicting streamflow dynamics. The LMFF model was able to improve the average of root mean square error (RMSE_ave) and average of mean absolute percentage error (MAPE_ave) values of the multiple linear regression forecasts by about 18% and 21%, respectively. In spite of the fact that the RBF model acted better for predicting the highest range of flow rate (flood events, RMSE_ave/RBF = 26.8 m³/s vs. RMSE_ave/LMFF = 40.2 m³/s), in general, the results suggested that the LMFF method was somehow superior to the RBF method in predicting watershed runoff (RMSE/LMFF = 18.8 m³/s vs. RMSE/RBF = 19.2 m³/s). Eventually, statistical differences between measured and predicted medians were evaluated using Mann-Whitney test, and differences in variances were evaluated using the Levene's test.     

快速二阶BP网络及其在城市用水量预测中的应用

针对BP网络收敛速度慢,易导致局部极小值的缺点,提出一种快速二阶BP网络,并以城市年用水量预测为例,与BP网络对比,结果表明,该方法加快了收敛速度,提出了结果的准确度。     

Identification of process disturbance using SPC/EPC and neural networks

Since solely using statistical process control (SPC) and engineering process control (EPC) cannot optimally control the manufacturing process, lots of studies have been devoted to the integrated use of SPC and EPC. The majority of these studies have reported that the integrated approach has better performance than that by using only SPC or EPC. Almost all these studies have assumed that the assignable causes of process disturbance can be identified and removed by SPC techniques. However, these techniques are typically time-consuming and thus make the search hard to implement in practice. In this paper, the EPC and neural network scheme were integrated in identifying the assignable causes of the underlying disturbance. For finding the appropriate setup of the networks' parameters, such as the number of hidden nodes and the suitable input variables, the all-possible-regression selection procedure is applied. For comparison, two SPC charts, Shewhart and cumulative sum (Cusum) charts were also developed for the same data sets. As the results reveal, the proposed approaches outperform the other methods and the shift of disturbance can be identified successfully.     

PSO优化的神经网络在教学质量评价中的应用

针对以往教学质量评估体系中存在的问题,利用粒子群优化算法(PSO)训练的神经网络建立教学质量评估数学模型.该方法使用由PSO训练的BP模型来拟合影响教师教学质量评价的众多指标与评价结果之间的复杂关系.实验结果表明,运用人工神经网络能更好的建立综合评价系统,用于满足更多范围的系统综合评价.     

Bayesian selective combination of multiple neural networks for improving long-range predictions in nonlinear process modelling

A Bayesian selective combination method is proposed for combining multiple neural networks in nonlinear dynamic process modelling. Instead of using fixed combination weights, the probability of a particular network being the true model is used as the combination weight for combining that network. The prior probability is calculated using the sum of squared errors of individual networks on a sliding window covering the most recent sampling times. A nearest neighbour method is used for estimating the network error for a given input data point, which is then used in calculating the combination weights for individual networks. Forward selection and backward elimination are used to select the individual networks to be combined. In forward selection, individual networks are gradually added into the aggregated network until the aggregated network error on the original training and testing data sets cannot be further reduced. In backward elimination, all the individual networks are initially aggregated and some of the individual networks are then gradually eliminated until the aggregated network error on the original training and testing data sets cannot be further reduced. Application results demonstrate that the proposed techniques can significantly improve model generalisation and perform better than aggregating all the individual networks.     

Application of artificial neural networks for the development of a signal monitoring system

A prototype of a Signal Monitoring System (SMS) utilizing artificial neural networks is developed in this work. The prototype system is unique in: 1) its utilization of state-of-the-art technology in pattern recognition such as the Adaptive Resonance Theory family of neural networks, and 2) the integration of neural network results of pattern recognition and fault identification databases.
The system is developed in an X-windows environment that offers an excellent Graphical User Interface (GUI). Motif software is used to build the GUI. The system is user-friendly, menu-driven, and allows the user to select signals and paradigms of interest. The system provides the status or condition of the signals tested as either normal or faulty. In the case of faulty status, SMS, through an integrated database, identifies the fault and indicates the progress of the fault relative to the normal condition as well as relative to the previous tests.
Nuclear reactor signals from an Experimental Breeder Reactor are analyzed to closely represent actual reactor operational data. The signals are both measured signals collected by a Data Acquisition System as well as simulated signals.     

The essential order of approximation for nearly exponential type neural networks

1 Introduction Artificial neural networks have been extensively applied in various fields of science and engineering. Why is so is mainly because the feedforward neural networks (FNNs) have the universal approximation capability[1-9]. A typical example of…