This paper defines a form of structural learning for Bayesian classifiers using the heuristic Meta Optimization based on Variable Meshes, VMO. The method is based on finding the topology of arbitrary Bayesian networks that best classify the data. In the process, a wraparound technique is used for the supervised learning task. This optimization problem is complex, since space grows exponentially depending on the number of variables. The proposal was tested in an educational setting and compared with other Bayesian classifiers using free access software such as Elvira and Weca.
Real Time Impact Factor:
Pending
Author Name: Byron Oviedo Bayas,*, Erika Zamora Cevallos, Amilkar Puris Cáceres, Cristian Zambrano Vega, Jorge Gómez Gómez
URL: View PDF
Keywords: Bayesian networks, structural training, heuristic goal, variable mesh, optimization.
ISSN: ISSN Impreso: 1316-4821
EISSN: ISNN Digital: 2542-3401
EOI/DOI:
Add Citation
Views: 1
