A machine learning framework in which two neural networks contest with each other in the form of a zero-sum game, where one agent’s gain is another agent’s loss. A GAN learns to generate new data with the same statistics as the training set.
Sources:
NIST AI 100-2e2025