Genomic data analysis using Bayesian networks with contextual dependencies

OData support
Supervisor:
Dr. Antal Péter
Department of Measurement and Information Systems

The Bayesian networks are one of the most important methods in representing uncertain knowledge. The advantage of using Bayesian nets is that it is capable of representing multivariant distributions, the model explicitly contains the independencies between variables. The Bayesian nets can be constructed by using a priori knowledge of experts, or by learning from statistical data.

One big problem with the latest approach is that it needs a huge amount of statistical data, therefor a lot of research focuses on probability modells. The thesis examines Bayesian networks, which has local models based on contextual independence.

Three different kind of local model will be implemented: the default table, the decision tree and the decision graph. We are trying to create more accurate models by using them when learning from statistical data.

Downloads

Please sign in to download the files of this thesis.