Applying Neural Network Approach with Imperialist Competitive Algorithm for Software Reliability Prediction

Abstract = 0 times | PDF = 9 times


Shirin Noekhah Naomie binti Salim


With the presence of software systems in variety critical field, the complexity level of software has increased, so software reliability has become an important issue and more difficult to guarantee. Software reliability is one of the significant factors of software quality applied to evaluate to what extend the desired software is trustable. To overcome the problem of dependency to human power and time limitation for software reliability prediction, researchers have focused on soft computing approaches. Nevertheless, the new soft computing techniques such as Neural Network and Fuzzy Logic have some problems, like no solid mathematical foundation for analysis, trap in local minima and convergence problem. This paper proposed a model to predict software reliability by hybridizing the Multi-Layer Perceptron Neural Network (MLP) and Imperialist Competitive Algorithm (ICA). This model has solved most of the previous problems, such as the convergence problem, requiring a large amount of data, and it can be applied in complex software systems. Numerical results show that both training and testing stages of the proposed model outperform existing approaches in terms of predicting the number of software failures.


Soft computing, reliability of software, Multi-Layer Perceptron Neural Network, Imperialist Competitive Algorithm