Generative Adversarial Networks (GANs) have revolutionized the field of artificial intelligence by enabling the generation of realistic data. In my bachelor thesis, titled "Generate fake data with Generative Adversarial Networks (GANs)," conducted under the supervision of Dr. MohammadAli Khosravifard, I delved into the world of GAN algorithms. This article will provide an overview of my thesis, highlighting the key GAN models I explored and the practical applications demonstrated using TensorFlow.
GANs are composed of two neural networks competing against each other: the generator and the discriminator. The generator creates synthetic data samples, while the discriminator's role is to distinguish between real and fake samples. Through an iterative training process, these networks learn from each other, with the generator continuously improving its ability to generate realistic data, while the discriminator becomes more adept at identifying fake data.
To demonstrate the practical application of GAN algorithms, I implemented the reviewed models using the TensorFlow framework. Specifically, I focused on generating face images and handwritten digits. By training the GAN models on appropriate datasets, I witnessed the remarkable capability of GANs to generate synthetic data that closely resembles real images. This practical implementation allowed me to witness firsthand the power of GANs in generating realistic data.
