FengHZ / BlogComments

MIT License
0 stars 0 forks source link

A simple explaination of wasserstein distance #13

Closed FengHZ closed 4 years ago

FengHZ commented 5 years ago

https://fenghz.github.io/Wasserstein-distance/

FengHZ commented 5 years ago

Here you can leave a comment

eyuansu62 commented 5 years ago

您好,你公式里的#是啥意思呀

FengHZ commented 5 years ago

表示由T诱导的对度量空间的变换,对于变换T,诱导度量空间的变换 $T#$ 满足 $T#(u)(E)=u(T^{-1}(E))$

FengHZ commented 5 years ago

对于正态分布,p=2的时候可解,不是欧氏距离,要考虑方差的

eyuansu62 commented 5 years ago

那请问d纬度指的是什么呢?

---Original--- From: "FengHZ"<notifications@github.com> Date: Thu, Nov 14, 2019 22:19 PM To: "FengHZ/BlogComments"<BlogComments@noreply.github.com>; Cc: "Comment"<comment@noreply.github.com>;"eyuansu62"<772468951@qq.com>; Subject: Re: [FengHZ/BlogComments] A simple explaination of wasserstein distance (#13)

对于正态分布,p=2的时候可解,不是欧氏距离,要考虑方差的

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

FengHZ commented 5 years ago

应该就是R^n的n,这里显示错误了,我修改一下

eyuansu62 commented 5 years ago

您好, 当d=1时,F−1(z)−G−1(z) 这里的F是怎么得到的呀