International Journal of Application or Innovation in Engineering & Management (IJAIEM)
There is a major concern which has come across all the researchers in the field of software-reliability that is how to develop a universal model which can predict the reliability under all circumstances, since reliability-growth models predictions varies across different testing phases. The best way is to develop a model which makes no assumptions about the behavior of software failure. This is where neural network model comes into play. Neural network models have significant advantage over simple parametric reliability models because they only require failure history as input and make no assumptions.