Introduction to the Bayesian approach to linear regression. We address the question of why a Bayesian approach is preferable to using the MLE or MAP estimate.
Hi there. Thanks for the videos mathematicalmonk; they are very informative. I watched the ML10 series and the only thing that bugs me is the assumption that a,b are known. I have multivariate data which I want to model with a regression model and I am interested not only in finding yi but also the variance. Can someone please direct me to a tutorial or is there a way to infer the above values?
I think u are right. Here monk didn't explain very clearly. In fact, MLE can also give us P(y|x, D). And in MLE, we choose our basis function by optimizing the loss function, which he mentioned in previous videos. So, I think the reason why we choose Bayesian is not so correct in this video.
This vid makes references to watching other vids before proceeding to the next one !
Amazing explanation! Save my ML homework, thank you so much
Hi there. Thanks for the videos mathematicalmonk; they are very informative. I watched the ML10 series and the only thing that bugs me is the assumption that a,b are known. I have multivariate data which I want to model with a regression model and I am interested not only in finding yi but also the variance. Can someone please direct me to a tutorial or is there a way to infer the above values?
Great stuff, thanks for the effort!
When reached 5:44, I have a question: in MLE, we assumed $Y$ follows Gaussian distribution right? this is not enough?
I think u are right. Here monk didn't explain very clearly. In fact, MLE can also give us P(y|x, D). And in MLE, we choose our basis function by optimizing the loss function, which he mentioned in previous videos. So, I think the reason why we choose Bayesian is not so correct in this video.
great thanks. I am quite new in this field. May I know why all w have mean of zero? It seems to me these coefficients should have non zero means.
我也不明白，😔
5 years passed by and no one knows the answer ;(
@@keno2055 This video was made over 10 years ago and has very few views. That's probably why.
Making prior assumprion about the distribution of theta being standard normal N(0,1). Common prior assumption
How can we assume a,b are known?
Damnit, your videos rule
excellent! dont sell bayesian models short though! they are good for many more reasons :)
How do the MLE estimates represent our uncertainty?
it doesn't. bayesian does.