Core Concepts

What Is a GAN (Generative Adversarial Network)?

A generative adversarial network (GAN), introduced in 2014, trains two neural networks against each other: a generator that creates fake data and a discriminator that tries to tell fakes from real. This competition drives the generator to produce increasingly realistic output. GANs powered early photorealistic image generation and deepfakes before diffusion models rose to prominence.