Lifted Temporal Most Probable Explanation

摘要

The lifted dynamic junction tree algorithm (LDJT) answers filtering and prediction queries efficiently for temporal probabilistic relational models by building and then reusing a first-order cluster representation of a knowledge base for multiple queries and time steps. Another type of query asks for a most probable explanation (MPE) for given events. Specifically, this paper contributes (i) LDJT~(mpe) to efficiently solve the temporal MPE problem for temporal probabilistic relational models and (ii) a combination of LDJT and LDJT~(mpe) to efficiently answer assignment queries for a given number of time steps.