Machine Learning and Applications, Fourth International Conference on (2009)
Miami Beach, Florida
Dec. 13, 2009 to Dec. 15, 2009
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/ICMLA.2009.78
The General Solution DAGs (GS-DAGs) are a method used to solve the Unconstrained Influence Diagrams (UIDs). In the first part of this paper, we determine the strict upper bound on the size of the GS-DAG with respect to the size of the UID being solved. In the second part, we introduce a new type of GS-DAG multinode, which reduces the number of edges of the GS-DAGs significantly. This has a huge impact on the evaluation of the GS-DAGs, because the number of edges heavily influences the number of computed and stored potential tables. The results presented in this article are also of a great importance from the point of view of approximation methods, which appeared recently and try to outperform the GS-DAG approach with an unknown complexity (until now).
unconstrained influence diagrams, decision graphs, exact solution, complexity
Z. Reitermanová, O. Sýkora and J. Iša, "On the Complexity of General Solution DAGs," Machine Learning and Applications, Fourth International Conference on(ICMLA), Miami Beach, Florida, 2009, pp. 673-678.