[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.
1705.02438] Face Super-Resolution Through Wasserstein GANs
Wasserstein GAN
GitHub - dhyaaalayed/wgan-gaussian: An implementation of Wasserstein GAN to generate 5 different Gaussian distributions
Applied Sciences, Free Full-Text
GANs in computer vision - Improved training with Wasserstein distance, game theory control and progressively growing schemes
CS 182: Lecture 19: Part 3: GANs
WGAN-GP Loss Explained
WGAN: Wasserstein Generative Adversarial Networks
Advancements in GAN Discriminator Models: From Vanishing Gradient to WGAN-GP – CryptLabs
PDF] From GAN to WGAN
4. Generative Adversarial Networks - Generative Deep Learning [Book]
What is the practical difference between the traditional GAN formulation and the Wasserstein GAN? - Quora
How can I calculate this steadily decreasing WGAN metric? - PyTorch Forums
python - Wasserstein loss can be negative? - Stack Overflow