Generative Adversarial Network GAN

Two networks trained against each other — one generating fakes, one detecting them — until the fakes are convincing.

Generative AI Advanced 1 min read

Also known as: GAN

Definition

A GAN pits a generator, which produces samples from noise, against a discriminator, which tries to tell real from generated. Each improves by defeating the other, and at equilibrium the generator produces samples the discriminator cannot distinguish.

GANs dominated image generation for several years and produce sharp results quickly. They are also notoriously unstable to train, and prone to mode collapse — producing a narrow slice of the data distribution very convincingly. Diffusion models displaced them largely on those two counts.

How it relates to other terms

Contrast with Diffusion Model A generative model that learns to reverse a gradual noising process, turning random noise into an image step b...
Referenced by
Diffusion Model Contrast with

Related terms

Prompt Engineering

Writing the input to a model deliberately — with context, examples and constraints — to get a more reliable output.

Beginner Generative AI

Temperature

A setting that controls how much randomness goes into choosing each token — low is predictable, high is varied.

Beginner Generative AI

Diffusion Model

A generative model that learns to reverse a gradual noising process, turning random noise into an image step by step.

Advanced Generative AI

Was this definition helpful?

Definitions are reviewed by our editorial team. Spotted a problem? Tell us.

Keep reading

Browse the glossary Suggest a correction