Abstract :
In this paper, we propose the use of wavelet neural networks (WNN) to predict software reliability. In WNN, we employed two kinds of wavelets - Morlet wavelet and Gaussian wavelet as transfer functions resulting in two variants of WNN. The effectiveness of WNN is demonstrated on a data set taken from literature. Its performance is compared with that of multiple linear regression (MLR), multivariate adaptive regression splines (MARS), backpropagation trained neural network (BPNN), threshold accepting trained neural network (TANN), pi-sigma network (PSN), general regression neural network (GRNN), dynamic evolving neuro-fuzzy inference system (DENFIS) and TreeNet in terms of normalized root mean square error (NRMSE) obtained on test data. Based on the experiments performed, it is observed that the WNN outperformed all the other techniques.
Keywords :
backpropagation; fuzzy reasoning; mean square error methods; neural nets; regression analysis; software reliability; wavelet transforms; Gaussian wavelet; Morlet wavelet; TreeNet; backpropagation trained neural network; dynamic evolving neuro-fuzzy inference system; general regression neural network; multiple linear regression; multivariate adaptive regression splines; normalized root mean square error; pi-sigma network; software reliability prediction; threshold accepting trained neural network; wavelet neural networks; Application software; Bayesian methods; Computational intelligence; Fourier transforms; Mars; Neural networks; Predictive models; Regression tree analysis; Software reliability; Support vector machines;
