GANs are a class of generative models consisting of two neural networks—a generator and a discriminator—that are trained simultaneously in a minimax game. The generator tries to produce data that resembles the real distribution (e.g., realistic images), while the discriminator learns to distinguish between real and generated samples.

The generator improves by learning to fool the discriminator, and the discriminator improves by getting better at spotting fakes. Over time, the generator ideally produces high-quality, realistic outputs.

GANs have been used for image synthesis, super-resolution, style transfer, data augmentation, and even audio and video generation. Variants like DCGAN, StyleGAN, and CycleGAN improve stability or adapt to different tasks (e.g., unpaired image translation).

Training GANs is challenging due to issues like mode collapse (generator produces limited variety), vanishing gradients, and instability. Techniques like feature matching, Wasserstein loss, and spectral normalization help mitigate these problems.