Software quality assurance for object-oriented systems using meta-heuristic search techniques

Abstract

Identifying fault prone modules at the very early stage of software development life cycle is very much necessary. This helps software developers to concentrate more on quality assurance, use the man power in proper perspective and mainly reduce the fault removal cost to be in-cured for the software system being developed. In literature, it is found that numerous authors have come up with cost based evaluation frameworks to find the effectiveness of the proposed fault prediction model based on the application of neural network models, but less emphasizes is provided on the manner in which networks are trained, where in a elementary approach of assigning random weights to the nodes is followed. This paper evaluates the capability of meta-heuristic search techniques in software fault classification for Apache Integration Framework. Genetic algorithm (GA) and Particle swarm optimization (PSO) is coupled with neural network for designing prediction models. It is observed that Modified Neuro PSO model is more effective and efficient in classifying faults accurately when compared to Neuro-GA, Adaptive Neuro-GA and Neuro-PSO.