Merging Evolutionary Approach with Neural Network for automatic Creation and Detection of faults in test cases

Software testing is an important activity of the software development process, because most of the development cost is spent in the testing phase. In order to test the software, it is necessary to write test cases. Manually creating test data is very time consuming, especially for complex problems. It consumes resource and time. In our study we use evolutionary approach. The evolutionary approach that is used here is Genetic Algorithm. It automatically creates test cases for software application. For this, it first selects important and best test cases among all the test cases, on the basis of fitness function who will decide which test cases are good or best for execution. Then these selected test cases are used to create new test cases. But test cases are more likely to have faults. For this, multi-layer neural network is first trained by using randomly generated test data that confirm to the specification. Once training is over, a comparison tool, fault predictor is used for detecting the presence of fault in the software application. It compares the output of the application under test with the output of trained neural network and makes the decision about the presence of fault in the software application