Generative Adversarial Network (GAN)
A generative architecture with two competing networks: a generator that creates synthetic data and a discriminator that distinguishes real from generated data. Training them in competition produces increasingly realistic outputs.