GANs, Explained Simply

A creative duel

A GAN pits two networks against each other: a generator invents fake samples, and a discriminator tries to tell fake from real. As each improves, the generator learns to produce increasingly convincing data.

What they did

GANs powered an era of realistic face generation and image-to-image translation, and they remain useful for super-resolution and data augmentation.

GANs vs. diffusion

Diffusion models have largely overtaken GANs for image generation thanks to more stable training and higher quality. But the adversarial idea — learning by competition — remains an elegant and influential concept.