Generative Adversarial Networks for Code Testing

The goal of this project is investigate the use of Generative Adversarial Networks (GAN) to generate automate code tests for software applications. The scope of the project includes researching relevant research studies and background information, design and implement new GAN models to generate unit test code, conduct experimental analysis and evaluation of the designed models and compare the results. Relevant dataset should be also generated (consisting of simple source codes and their unit test code counterparts) so it can be used in the evaluation. The project would also involve designing a simple UI that allow to use the developed GAN models.