案例库 · 研发与科研 · 技术决策 · 2014–2018
这条还没译成中文，下面是英文原文。
Generative adversarial networks made two models train against each other.
Goodfellow's GANs pit a generator against a discriminator, so each improves by trying to beat the other.
OpenAI · Google Brain
那一手
Generative models, which create new data from a distribution, were often hard to train or produced blurry output because they required complicated inference.
Goodfellow and colleagues proposed pairing a generator with a discriminator in a minimax game; the generator learns by trying to fool the discriminator, and the discriminator learns by resisting.
The framework required no Markov chains or approximate inference and could be trained with standard backpropagation, and it produced convincing samples.
为什么管用
- The discriminator provides a learning signal without a labelled target
- The adversarial objective is a scoreboard that improves automatically
- Backpropagation-only training makes the approach practical
- Competition between the two networks drives both toward better performance
可以搬走什么
You don't always need an objective truth to train against: a well-designed adversary can be the scoreboard, so two imperfect systems push each other toward competence.
后来呢
GANs became one of the most influential ideas in machine learning, powering image generation, style transfer and synthetic-media research. They also raised concerns about deepfakes and misuse, and a large body of follow-up work stabilised and extended the original training objective.
资料来源
发现哪里写错了？告诉我们。