wk-7

RUNNING HEAD: NETWORKS 1

NETWORKS 2

Naïve Bayes and Bayesian Networks

Student’s Name

Institutional Affiliation

Date

A Bayesian network is a probability graphical model that is used to represent a set of variables which are in nodes (Sharda, et al, 2020). Their conditional dependencies are also accounted for in this model. Naïve Bayes network is a technique that is used to classify and assign class labels to an available set of variables. This model of classification usually assumes the independence of each feature and no correlation or relationship is usually considered between the features. Naïve Bayes network is closely related to Bayesian network in the functionality of the two models.

Bayesian network uses a graph of nodes and models to represents a set of variables as the dependencies between these nodes as edges. This networks therefore makes inherent assumptions on the independence and dependence of the variables while in real world, these variables are never entirely independent (Alhussan, & El Hindi, 2016). For practical usage of the Bayesian model however, all the variables that are nearly independent are completely independent. Naïve Bayes model makes simplified assumptions that the given class labels are independent of each other. This independence assumption works well in most pf the cases even though in reality, there is no independence of the variables.

Batesian networks are Directed Acyclic Graphs (DAGs) whereby the nodes represent the variables such as hypotheses and other observable quantities. Edges represent the conditional dependencies and these are the nodes that are not connected. In developing the Bayesian networks therefore, begins with the creation of DAG (Sharda, et al, 2020). The G is constructed in a way that X satisfies the Markov property; set of nodes consisting parents, and children. There is assessment of conditional probability distributions of each variable. In specific cases when the variables are discrete, then the distribution of the X is the product of the distributions and in this case, the Bayesian network in respect to G.

References

Alhussan, A., & El Hindi, K. (2016). Selectively fine-tuning Bayesian network learning algorithm. International Journal of Pattern Recognition and Artificial Intelligence, 30(08), 1651005. Retrieved from https://www.worldscientific.com/doi/abs/10.1142/S0218001416510058

Sharda, R., Delen, D., Turban, E. (2020). Analytics, Data Science, & Artificial Intelligence: Systems for Decision Support 11E. ISBN: 978-0-13-519201-6.