Posterior Probability Calculations

The engine.
Post Reply
Posts: 3
Joined: Wed Jan 27, 2021 2:36 pm

Posterior Probability Calculations

Post by pradhanojas » Mon Sep 13, 2021 3:18 pm


How are the posterior probabilities calculated in the SMILE engine? Is the inference considered to be a Kalman filter or a Particle filter? I would like to use dynamic BNs for future predictions and wanted to know how I could do that using the DBN structure created using the SMILE engine. Any help would be highly appreciated.

marek [BayesFusion]
Site Admin
Posts: 338
Joined: Tue Dec 11, 2007 4:24 pm

Re: Posterior Probability Calculations

Post by marek [BayesFusion] » Mon Sep 13, 2021 9:02 pm

SMILE unrolls a DBN and treats it as a static BN in which each of the time steps is represented explicitly. In this sense, given observations of any variables in any time steps, it calculates the conditional posterior probability distribution over the remaining variables. I believe that this procedure is more general than Kalman filter (used for linear or linearized processes) and particle filter, which are Monte Carlo algorithms. SMILE bases its solution on whatever algorithm you select as the default algorithm and that includes the (exact!) clustering algorithm.
I hope this helps.


Post Reply