I have an issue with linear Gaussian models which have low variance. If the variance of the linear Gaussian is too low then means of the variables in the joint tend to zero. I have attached a simple MATLAB script showing this:
means_issue.txt
This may be a problem for my application since I have these linear Gaussian models in a dynamic Bayesian network whos time t belief updates after each time step, possibly reducing the variances of multiple nodes.
What is the recommended way to work-around this issue?
Hi,
I have an issue with linear Gaussian models which have low variance. If the variance of the linear Gaussian is too low then means of the variables in the joint tend to zero. I have attached a simple MATLAB script showing this: means_issue.txt
This may be a problem for my application since I have these linear Gaussian models in a dynamic Bayesian network whos time t belief updates after each time step, possibly reducing the variances of multiple nodes.
What is the recommended way to work-around this issue?