I have a question about the formula in the paper

[applefl]

The proposed weighted GMM is：

but for two variable x1, x2 with different weight w1,w2,
if x1=x2, w1>w2, why the weighted GMM p(x1,w1)<p(x2,w2)

[isrish]

x1, x2 are non-iid samples ( independent , non-identically distributed samples) so you can't compare as you wrote. Rather compare the distribution x1 ~ N(mu, sigma/w1) and x2~N(mu, sigma/w2)

…

On Fri, Sep 25, 2020 at 9:08 PM applefl ***@***.***> wrote: The proposed weighted GMM is： [image: image] <https://user-images.githubusercontent.com/12521662/94326520-e5b2b580-ffd6-11ea-8b60-0371dc2f7f97.png> but for two variable x1, x2 with different weight w1,w2, if x1=x2, w1>w2, why the weighted GMM p(x1,w1)<p(x2,w2) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAWND323VD4GCK7LFSAENEDSHU5JJANCNFSM4R2R5MGQ> .

[applefl]

Thanks a lot, I think I should go to understand the Gaussian mixture model again.