-
## 一言でいうと
VAEやGANなどの生成系のタスクでは、「真の分布」との距離の最小化を目的にしている。つまり「距離」の定義はモデルの精度の大きな要素で、GANではこの自由度が高い反面学習が安定しない要因になっていた。そこでWasserstein距離を使うと勾配が消失せず学習が安定したという話。
### 論文リンク
https://arxiv.org/abs/1701.0787…
-
https://arxiv.org/pdf/1701.07875.pdf
We introduce a new algorithm named WGAN, an alternative to traditional GAN training. In this new model, we show that we can improve the stability of learning, g…
leo-p updated
7 years ago
-
# Wasserstein GAN #
- Author: Martin Arjovsky, Soumith Chintala, Léon Bottou
- Origin: [http://www.gitxiv.com/posts/BLGXoByEcDgGF3rxx/wasserstein-gan](http://www.gitxiv.com/posts/BLGXoByEcDgGF3rxx…
-
**Additional context**
I would like to add a WGAN model trained on Celeb A dataset in pytorch.
-
Hi,
I'm now studying your paper "Max-Sliced Wasserstein Distance and its use for GANs". I want to study the supplementary material, but I cannot find it on the net. Would you please give me a addr…
-
Hi, I'm trying to apply spectral normalization to wasserstein gans. I've failed to make it work in my project so tried your repository to get more intuition of how to train them. However I had no trai…
-
Hello,
Thank you for the implementation of the Wasserstein GANs mode and GP loss!
I followed the way proposed here: https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix/issues/439
It works fo…
-
Dear author,
Thanks for your friendly implementation of popular GANs. I am reading your implementation of [Wasserstein GANs](https://github.com/znxlwm/pytorch-generative-model-collections/blob/mast…
-
## 一言でいうと
Maxmarginの手法を抽象化して、SVM/GANを同じ枠組みでとらえようという研究。確率分布の距離を使用するGANが
学習効果にすぐれることを示し、発展させた手法としてL∞ gradient normを使用する方法を提案している。
### 論文リンク
https://arxiv.org/abs/1910.06922
### 著者/所属機関
A…
-
Have you tried WGAN with GAN-CLS? I am trying to use use Wasserstein distance, in GAN CLS and I modified the cost and clipped the weights, but it GAN doesn't learn anything. If I don't clip the weight…