Explains GAN fundamentals through a game-theory generator-vs-critic framing
Covers why GANs are hard to train: instability and convergence failure
Reference primer available in Korean and on arXiv
Key quotes
“It is inspired by game theory: two models, a generator and a critic, are competing with each other while making each other stronger at the same time.”
“However, it is rather challenging to train a GAN model, as people are facing issues like training instability or failure to converge.”
Transcript
[Updated on 2018-09-30: thanks to Yoonju, we have this post translated in Korean !] [Updated on 2019-04-18: this post is also available on arXiv .] Generative adversarial network (GAN) has shown great results in many generative tasks to replicate the real-world rich content such as images, human language, and music. It is inspired by game theory: two models, a generator and a critic, are competing with each other while making each other stronger at the same time. However, it is rather challenging to train a GAN model, as people are facing issues like training instability or failure to converge.
