What Is GAN (Generative Adversarial Network)?
A GAN (Generative Adversarial Network) is a deep learning architecture consisting of two neural networks — a generator that creates synthetic data and a discriminator that evaluates whether data is real or generated — that are trained simultaneously in a competitive process to produce increasingly realistic outputs.
How GAN (Generative Adversarial Network) Works
GANs work through an adversarial game: the generator tries to create data realistic enough to fool the discriminator, while the discriminator tries to distinguish real data from fakes. As training progresses, both networks improve — the generator produces increasingly convincing outputs, and the discriminator becomes more discerning. This competition drives the generator toward creating highly realistic data. GANs were the dominant image generation technique before diffusion models and are still used for tasks like super-resolution, image-to-image translation, and data augmentation. StyleGAN, for example, can generate photorealistic faces of people who do not exist.
Real-World Examples
StyleGAN generating photorealistic portraits of fictional people for the website 'This Person Does Not Exist'
A GAN converting daytime city photos to nighttime versions while preserving the scene structure
Researchers using GANs to generate synthetic medical images for training when real patient data is limited