This paper discusses Generative Adversarial networks, one of the latest techniques to solve the problem of data imbalance for software defect prediction tasks. When the data available is not enough in a task to frame a machine learning model, it is difficult to construct an accurate model. Generally Software engineering activities like defect prediction, effort estimation etc., were done on data available from open source datasets which had less data. The software defect prediction data available are not only smaller in size, but also imbalanced in nature with very less data found in the defective class. In order to overcome this data imbalance, artificial data generation techniques have been employed. In this work we try to improve the software defect prediction performance in projects, where the data available is less and imbalanced, using Generative Adversarial Networks (GANs).