EM.4: Gaussian mixture model (GMM)

[@aayushsaxena7374]

best tutorial for GMM on KZbin :)

[@FelixSchober]

Great video. Very easy to understand!

[@sanjaykrish8719]

Why do we need to model the data came from different Gaussian sources? Can you tell a really world scenario where can be modeled as coming from different Gaussian sources

[@Milind08662]

What will the density function for 3d data

[@RomilAggarwal611]

I had 1 doubt. I want to use 2 layer GMM. What I mean by that is say I have n data points(of course not 1-d), I first calculate the GMM for each data point and then I want to combine all the learned GMMs and form a single GMM. What is the best approach to do this (I have calculated the individual GMMs, I'm facing a problem in combining them)?

[@bharathramagoni8188]

Nice. Very nice . Helped me a lot. Thanks

[@anu1594]

i have a doubt sir, its more of a confusion. what is C ? is it the source or the gaussian? and if C is a gaussian then what is attritube 'a' and 'b' for ? what do they represent , cant quite relate it to your previous lectures on 1-D could you please help? thank you

[@damirrakhimov7320]

Thank you a lot! Very usefull video set.

[@AdrianVrabie]

Victor! :) when u use the bayesian rule to calculate P(c|x_i), where do u get the P(c) from? Didn't u just take two sources at random from the very beginning... and u never mention them in the video.

[@yusufazishennytriyudhantor9417]

I assume that at the beginning the prior of data Xi coming from source Cp would be equal to 1/n with n equals to how many cluster we want get. But than at the next iteration we update the prior by averaging the probabilty of how likely data Xi coming from source Cp

[@cominup85]

Good explanation, thanks