Data Augmentation Using Generative Adversarial Network

Now a days, Deep Learning has made appreciable development which introduces intelligence in machines to work like human brain. For this learning, the presence of large and balanced dataset is essential so that we can train the machines more efficiently. However finding such data in real world is rare, and creating these data sets is a complex task. So Generative Ad- versarial Networks (GANs) are used to create dataset to enhance the unsupervised learning. In this proposed work, Convolutional GAN is used for data augmentation which produces more realistic datasets and then we analyse the performance of this GAN by doing hyper-parameter tuning of opitmizers and activation functions.