Generative adversarial network

[JEN-uh-ruh-tiv ad-vuh-SAIR-ee-uhl NET-wurk] /ˈdʒɛn(ə)rətɪv ˌadvəˈsɛːriəl ˈnɛtwəːk/

A machine learning framework where two deep neural networks, a generator and a discriminator, are pitted against each other to eventually train the generator network to create highly realistic synthetic data. Both models are continuously updated, with the generator attempting to mimic a real-world dataset, and the discriminator judging whether this incoming data is real (from the training dataset) or fake (generated by the other neural network). Once the generator network can consistently fool the discriminator, the generator neural network can be used to create highly realistic synthetic data in the form of images, sounds, text, or whatever other form of data it was trained on.