scientific article; zbMATH DE number 19874
[For the entire collection see Zbl 0741.68004.] Updating belief networks is exponential in the number of nodes. All practical implementations therefore restrict updating to a small subset of the entire system. A systematic and coherent method therefore demands an optimal splitting of the network, a problem intrinsically \(NP\)-hard. The author represents the network as a hypergraph and introduces an algorithm that minimizes the total number of states, i.e., represents the total effort required for a full updating of a system on \(N\) states by a minimal sum \(\Sigma 2^{mi}\) where \(\Sigma m_ i=N\).
- Tree decompositions with small cost
- All roads lead to Rome -- new search methods for the optimal triangulation problem
- A review on evolutionary algorithms in Bayesian network learning and inference tasks
- Operating with potentials of discrete variables
- Local conditioning in Bayesian networks
- A model for compressing probabilities in belief networks
- Creating non-minimal triangulations for use in inference in mixed stochastic/deterministic graphical models
- Decomposition of arithmetical np-hard problems
- A new algorithm for decomposition of graphical models
- An improved Hara-Takamura procedure by sharing computations on junction tree in Gaussian graphical models
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3979385)