[1] Geometry Score: A Method For Comparing Generative Adversarial Networks
Valentin Khrulkov, Ivan Oseledets
Skolkovo Institute of Science and Technology, Institute of Numerical Mathematics RAS
http://proceedings.mlr.press/v80/khrulkov18a/khrulkov18a.pdf
下面是原始数据和GAN生成的数据对比如下
计算RLT的算法伪代码如下
其中RLT为relative living times。
计算几何相似度的算法伪代码如下
不同数据集的平均RLT对比如下
不同方法的几何得分对比如下
代码地址
https://github.com/KhrulkovV/geometry-score
https://github.com/igul222/improved_wgan_training
https://github.com/carpedm20/DCGAN-tensorflow
https://github.com/hep-lbdl/CaloGAN
我是分割线
[2] K-Beam Minimax: Efficient Optimization for Deep Adversarial Learning
Jihun Hamm, Yung-Kyun Noh
The Ohio State