Historical contribution
What Goodfellow contributed
A generative adversarial network trains a generator and discriminator in competition, allowing models to learn how to produce data resembling a training distribution.
Why it mattered
GANs accelerated research in synthetic images and other generative media and clarified the power and instability of adversarial training.
Connected topics
Where this fits in ITEM
Evidence
Sources and further reading
- Generative Adversarial Nets
Ian Goodfellow · 2014