GAN

GAN (Generative Adversarial Network) is a generative machine learning architecture, introduced by Ian Goodfellow and colleagues in 2014, built from two neural networks locked in competition. The generator takes a random vector from latent space and tries to produce realistic fake samples (e.g. images); the discriminator is shown a mix of real and generated samples and tries to tell which is which. They train against each other in a minimax game — the generator gets better at fooling the discriminator while the discriminator gets better at catching it — until, ideally, the fakes are indistinguishable from real data. GANs powered an era of photorealistic face generation (StyleGAN) and image-to-image translation, but are notoriously unstable to train (mode collapse, non-convergence) and have largely been overtaken by diffusion models for general-purpose image generation, though they remain useful where fast, single-pass sampling matters.