Drexel dragonThe Math ForumDonate to the Math Forum

Search All of the Math Forum:

Views expressed in these public forums are not endorsed by Drexel University or The Math Forum.

Math Forum » Discussions » sci.math.* » sci.stat.edu.independent

Topic: Numerical Issues about Computation of log likelihood in High Dimension
Replies: 0  

Advanced Search

Back to Topic List Back to Topic List  

Posts: 1
Registered: 5/17/06
Numerical Issues about Computation of log likelihood in High Dimension
Posted: May 17, 2006 3:20 AM
  Click to see the message monospaced in plain text Plain Text   Click to reply to this topic Reply

Dear Community Members
I'd like to compute likelihood score sum of the clustering made by KMeans algorithm by assuming that each cluster is a gaussian distributed samples. When I compute the likelihood for a sample using the following negatif log-likelihood formula

log-Likelihood = (x-mu)'*SigmaInv*(x-mu) + (d/2)ln(2pi)
+ (1/2)ln |SigmaInv|

where x : sample vector, mu : cluster mean sample vector, d : dimension of the sample vectors, SigmaInv : inverse of the covariance matrix sigma, the determinant of the |SigmaInv| goes beyond the numerical limit of the computer, and gives infinity error since d is very high such as about 400. What can be done to overcome this numerical problem and to compute the likelihoods of the samples ?


Point your RSS reader here for a feed of the latest messages in this topic.

[Privacy Policy] [Terms of Use]

© The Math Forum 1994-2015. All Rights Reserved.