Generative Adversarial Networks (GANs): A type of neural network architecture composed of two networks, a generator and a discriminator, that are trained to compete against each other in order to generate new, synthetic data that is indistinguishable from real data.