Optimal Junction Trees
This work addresses efficiency and optimality challenges in probabilistic inference for researchers in graphical models, but it appears incremental as it builds on existing junction tree methods.
The paper tackles the problem of optimality in belief updating for networks by addressing triangulation and junction tree construction, presenting a simple algorithm for building an optimal junction tree from a triangulated network and arguing that exact local methods must either be less efficient or face equivalent optimality issues.
The paper deals with optimality issues in connection with updating beliefs in networks. We address two processes: triangulation and construction of junction trees. In the first part, we give a simple algorithm for constructing an optimal junction tree from a triangulated network. In the second part, we argue that any exact method based on local calculations must either be less efficient than the junction tree method, or it has an optimality problem equivalent to that of triangulation.