Given a Bayesian network and evidence for the values of a subset of the variables, a standard question is to compute the posterior distribution on the remaining variables. The Gibbs sampling technique gives a discrete time Markov chain which samples from this posterior distribution. Is there a continuous time Markov process which does the same thing?
Remember to vote up questions/answers you find interesting or helpful (requires 15 reputation points)
|
1
|
|
|
|
|
1
|
For any discrete time Markov chain, you can get a continuous time Markov chain with the same stationary measure: just perform the same transitions as the discrete time Markov chain, but at times given by a Poisson process. |
||
|
|
