PREDICTING SOFTWARE RELIABILITY WIH NEURAL NETWORK ENSEMBLES

摘要

Software reliability is important for quantitatively characterizing software quality and estimating the duration of software testing period. Although traditional parametric software reliability growth models (SRGMs) such as nonhomogeneous Poisson process (NHPP) models have been successfully utilized in practical software reliability engineering, no single such model can obtain accurate prediction for all cases. Non-parametric models like neural networks, on the other hand, have shown to be effective for predicting software reliability. In this paper, we apply neural network ensembles for non-parametric software reliability prediction. The proposed method is compared with the single neural network and three parametric NHPP models. The experimental results prove that neural network ensembles are promising for software reliability prediction.