What is GAN (Generative Adversarial Network)?
TL;DR
A technique that generates realistic data by pitting two neural networks against each other. A foundational milestone in AI image generation.
GAN (Generative Adversarial Network): Definition & Explanation
A GAN (Generative Adversarial Network) is a type of generative model proposed by Ian Goodfellow and colleagues in 2014, consisting of two neural networks — a Generator and a Discriminator — that compete against each other to produce realistic data. The Generator tries to create data that looks real, while the Discriminator tries to distinguish real data from fakes. This 'adversarial' training process improves both networks over time. Many derivative models have been developed, including StyleGAN (high-quality face generation), CycleGAN (image style transfer), and Pix2Pix (image translation). While GANs laid the groundwork for the development of image generation AI, diffusion models have largely taken over as the dominant approach. GANs have also raised ethical concerns due to their potential misuse in deepfakes.