Software testing cost reduction with genetic algorithms and neural networks.

Highly complex and interconnected systems may suffer from intermittent or transient software failures. These are particularly difficult to diagnose without large quantities of test cases. This research focuses on a hybrid method for generating test cases. A genetic algorithm is first used to automatically generating large numbers of test cases to form a comprehensive test suite. These test suites are then used to train a neural network for regression testing and test suite augmentation. The results indicate that the genetic algorithm can produce a balanced test suite that, when combined with a neural network, can reduce the costs of software testing by reducing system run-time and human interaction