International Journal of Computational Intelligence and Applications

Software effort estimation is one of the most important tasks in software project management. Of several techniques suggested for estimating software development effort, the analogy-based reasoning, or Case-Based Reasoning (CBR), approaches stand out as promising techniques. In this paper, the benefits of using linguistic rather than numerical values in the analogy process for software effort estimation are investigated. The performance, in terms of accuracy and tolerance of imprecision, of two analogy-based software effort estimation models (Classical Analogy and Fuzzy Analogy, which use numerical and linguistic values respectively to describe software projects) is compared. Three research questions related to the performance of these two models are discussed and answered. This study uses the International Software Benchmarking Standards Group (ISBSG) dataset and confirms the usefulness of using linguistic instead of numerical values in analogy-based software effort estimation models.

This paper intends to compare various learning algorithms available for training the multi-layer perceptron (MLP) type of artificial neural networks (ANNs). By using different learning algorithms, this study investigates the performances of gradient descent (GD) algorithm; Levenberg-Marquardt (LM) algorithm; and also Boyden, Fletcher, Goldfarb and Shannon (BFGS) algorithm to predict the emissions of carbon dioxide (CO₂) in Malaysia. The impact factors of emissions, such as energy use; gross domestic product per capita; population density; combustible renewable and waste; also CO₂ intensity were employed in developing all ANN models investigated in this study. A wide variety of standard statistical performance evaluation measures were employed to evaluate the performances of various ANN models developed. The results obtained in this study indicate that the LM algorithm outperformed both BFGS and GD algorithms.

A dual-stage sampling (DSS) is proposed to reduce training samples of artificial neural network for establishing the analytical response surface model. The procedure of the DSS-based artificial neural network response surface (ANNRS) methodology is divided into two stages. At the rough stage, new sample is generated in the entire space by maximizing the minimum distance to existed samples. Then at the refine stage, new sample is produced in the same way, but is limited in the partial space between two samples with larger gradient and larger distance. And the efficiency of the DSS method is proved by a test function. Considering design variables about slot width and depth, permanent magnet (PM) width and height, and end tooth width, tubular permanent-magnet linear synchronous motor (TPMLSM) modeling problem is represented, whose thrust and ripple are calculated time-consumingly by finite element analysis. Then, The DSS-based ANNRS is applied to establish the TPMLSM model in good accuracy with fewer samples, lower consumption.

Electron beam butt welding of stainless steel (SS 304) and electrolytically tough pitched (ETP) copper plates was carried out according to central composite design of experiments. Three input parameters, namely accelerating voltage, beam current and weld speed were considered in the butt welding experiments of dissimilar metals. The weld-bead parameters, such as bead width and depth of penetration, and weld strength in terms of yield strength and ultimate tensile strength were measured as the responses of the process. Input-output relationships were established in the forward direction using regression analysis, back-propagation neural network (BPNN), genetic algorithm-tuned neural network (GANN) and particle swarm optimization algorithm-tuned neural network (PSONN). Reverse mapping of this process was also conducted using the BPNN, GANN and PSONN approaches, although the same could not be done from the obtained regression equations. Neural networks were found to tackle the problems of both forward and reverse mappings efficiently. However, neural networks tuned by the genetic algorithm and particle swarm optimization algorithm were seen to perform better than the BPNN in most of the cases but not all.

Fault tree analysis for nuclear power plant probabilistic safety assessment is an intricate process. Personal computer-based software systems have therefore been developed to conduct this analysis. However, all existing fault tree analysis software systems only accept quantitative data to characterized basic event reliabilities. In real-world applications, basic event reliabilities may not be represented by quantitative data but by qualitative justifications. The motivation of this work is to develop an intelligent system by fuzzy reliability algorithm in fault tree analysis, which can accept not only quantitative data but also qualitative information to characterized reliabilities of basic events. In this paper, a newly-developed system called InFaTAS-NuSA is presented and its main features and capabilities are discussed. To benchmark the applicability of the intelligent concept implemented in InFaTAS-NuSA, a case study is performed and the analysis results are compared to the results obtained from a well-known fault tree analysis software package. The results confirm that the intelligent concept implemented in InFaTAS-NuSA can be very useful to complement conventional fault tree analysis software systems.

One sub-field of Genetic Programming (GP) which has gained recent interest is semantic GP, in which programs are evolved by manipulating program semantics instead of program syntax. This paper introduces a new semantic GP algorithm, called SGP+, which is an extension of an existing algorithm called SGP. New crossover and mutation operators are introduced which address two of the major limitations of SGP: large program trees and reduced accuracy on high-arity problems. Experimental results on "deceptive" Boolean problems show that programs created by the SGP+ are 3.8 times smaller while still maintaining accuracy as good as, or better than, SGP. Additionally, a statistically significant improvement in program accuracy is observed for several high-arity Boolean problems.