Learning Decomposable Model-Structures by Applying Graphs of Minimal Connectors

S.-H. Kim and R.M. Kil (Korea)

Keywords

Bayes Network, Conditional Independence, Combined Structure.

Abstract

In AI and statistics, Bayesian networks and graphical models are very popular tools for modelling knowledge structures. Since model-learning is NP-hard, we need a new approach for learn ing a large model. We propose a method for learning the struc ture of a very large graphical model under the assumption that the models are decomposable.

Important Links:



Go Back