generative adversarial network

(computing) A system used in machine learning, consisting of two neural networks, one of which generates candidate solutions to a problem while the other evaluates and accepts or rejects them.