I am about to be finished with junction tree algorithm for inference in dynamic bayesian network. I am about to start with the map queries for inference.
Currently, the map queries finds the maximum value using the maximize operation. But for the dynamic bayesian network we need to compute a path that has the maximum probability also called the viterbi path.
The viterbi algorithm uses the famous dynamic programming algorithm paradigm, although it could be quite complicated to implement for a dynamic bayesian network.
Also, the inference in junction tree will further optimize the scale of operations, so the variable elimination will not try to lag the algorithm down.
I hope that the inference works well now.
Currently, the map queries finds the maximum value using the maximize operation. But for the dynamic bayesian network we need to compute a path that has the maximum probability also called the viterbi path.
The viterbi algorithm uses the famous dynamic programming algorithm paradigm, although it could be quite complicated to implement for a dynamic bayesian network.
Also, the inference in junction tree will further optimize the scale of operations, so the variable elimination will not try to lag the algorithm down.
I hope that the inference works well now.
No comments:
Post a Comment