Μπεϋζιανά δίκτυα : μέθοδοι και εφαρμογές
Bayesian networks : methods and applications

View/ Open
Keywords
Μπεϋζιανά Δίκτυα ; Κατευθυνόμενα γραφικά μοντέλα ; Κατευθυνόμενα ακυκλικά γραφήματα ; Markov Blanket algorithm ; Bayesian information criterionAbstract
Bayesian networks (BNs) offer a powerful framework for modelling and
inferring probabilistic relationships, which are prevalent in various industries
and scientific fields in the real world. This thesis delves into the theory
behind Bayesian networks, discussing methodologies for both structure and
parameter estimation. In addition, it extensively explores the features of the
bnlearn package, a main tool for inference in Bayesian networks in R, through
practical applications.
In addition, the thesis studies some applications from other sciences which
can help to better process results using code in R, thus exploring alternative
evaluation methods to mitigate the risk of selecting a network structure based
solely on score functions, which may favour overly complex structures.