Eitel Lauria - Marist College
Learning Structure and Parameters of Bayesian Networks
Bayesian networks are a natural combination between two areas in mathematics: graph theory and probability theory. A Bayesian Belief Network (BBN) encodes the probability distribution of a set of random variables by specifying a set of conditional independence assumptions together with a set of relationships among these variables and their related joint probabilities. A key feature of BBNs is that they enable us to model and reason about uncertainty, by providing a graphical representation that can help articulate expert beliefs about the dependencies between different variables. Also, they play an increasingly important role in the design and analysis of machine learning algorithms, constituting an innovative way of approaching problems related to Artificial Intelligence. This lecture addresses the issue of learning the structure and parameters of the Bayesian Networks from collected data. The research combines a) the use of heuristic methods to search the space of network structures; b) the analysis of a scoring function based on the Minimum Description Length Principle.
[Program Information | Faculty | Mathematics Seminar | Mathematics Links]
[Math Department Home Page | Marist College]