INVESTIGATION OF THE IMPACTS OF CONSTRAINT-BASED ALGORITHMS TO THE QUALITY OF BAYESIAN NETWORK STRUCTURE IN HYBRID ALGORITHMS FOR MEDICAL STUDIES

Abstract

The subject of how best to present the relationships between the variables under uncertainty was crucial in recent medical studies. In this study we compared the performance of hybrid structure learning algorithms with using different constraint based methods in skeleton construction phase. We used four different medical Bayesian networks for the comparison. In conclusion, we determined the most powerful combination of learning algorithms as a two phased hybrid method for building the structure.