Wu Ji

Application of maximum entropy principle to software failure prediction

Predicting failures from software input is still a tough issue. Two models, namely the surface model and structure model, are presented in this paper to predict failure by applying the maximum entropy principle. The surface model forecasts a failure from the statistical co-occurrence between input and failure, while the structure model does from the statistical cause-effect between fault and failure. To evaluate the models, precision is applied and 17 testing experiments are conducted on 5 programs. Based on the experiments, the surface model and structure model get an average precision of 0.876 and 0.858, respectively

A statistical model to locate faults at input levels

We present a statistical model to locate faults at the input level based on the failure patterns and the success patterns. The model neither needs to be fed with software module, code or trace information, nor does it require re-executing the program. To evaluate the model, precision and recall are adopted as the criteria. Five programs are examined and 17 testing experiments are conducted in which the model gains 0.803 in precision and 0.697 in recall on average.

Rolling Prediction Based Software Reliability Model Consideration with Learning Curve

Software reliability is an important factor for evaluating software quality in the domain of safety-critical software. The neural network prediction method has been widely used in reliability prediction area. However, Data noise and other issues make this approach easy to falling into local optimum, and reduce the accuracy of the prediction, it also affect the applicability of the model. In this paper, we consider the learning curve effect, and proposed a neural network based reliability prediction, utilize the rolling forecast method to elevate the accuracy and applicability of neural network. The method is validated through three groups of public data sets. And the results show a fairly accurate prediction capability.