Bayesian network learning with nominal, ordinal and continuous data

Mazuelos Wakabayashi, Manuel Takeru (2020). Bayesian network learning with nominal, ordinal and continuous data. Thesis (Master thesis), E.T.S. de Ingenieros Informáticos (UPM).

Description

Title: Bayesian network learning with nominal, ordinal and continuous data
Author/s:
  • Mazuelos Wakabayashi, Manuel Takeru
Contributor/s:
  • Fernández del Pozo de Salamanca, Juan Antonio
Item Type: Thesis (Master thesis)
Masters title: Inteligencia Artificial
Date: July 2020
Subjects:
Freetext Keywords: Bayesian Networks; Probability; Mutual Information; Parameter Learning; Structure Learning; Numerical estimation
Faculty: E.T.S. de Ingenieros Informáticos (UPM)
Department: Inteligencia Artificial
Creative Commons Licenses: Recognition - No derivative works - Non commercial

Full text

[img]
Preview
PDF - Requires a PDF viewer, such as GSview, Xpdf or Adobe Acrobat Reader
Download (3MB) | Preview

Abstract

Bayesian networks are a type of probabilistic graphic models composed of nodes and directed edges that serve to represent dependencies between the di erent random variables of the problem, as well as the conditional probability distributions associated to said variables. Originally, both the structure of the Network and the probability distributions were thought for discrete data, but due to the wide eld of application of Bayesian Networks, methods have been developed for other types of variables, as well as for problems with variables of various types. In this Master's Degree Dissertation, we seek to understand the operation of the di erent methods that have been developed to deal with the learning of Bayesian Networks with not only discrete variables, but also continuous, ordinal and mixed variables. Although one of the most common methods to treat Bayesian Networks with continuous variables is discretization, and some methods will be brie y explained, an attempt will be made to explain alternative methods to avoid the loss of information and the error with respect to the modelling of reality that it entails. After the completion of the thesis, it is expected to have an understanding of the subject, and to be able to transmit this knowledge to the reader in a clear and concise way. Structure learning methods will be explained, as well as the conditional probability distributions used to represent each kind of variable, together with the parameters to be learned. Also, methods of study of independence such as Mutual Information, and their estimation from a training database will be explained, as well as the Jonckheere-Terpstra Test for ordinal variables.

More information

Item ID: 63755
DC Identifier: http://oa.upm.es/63755/
OAI Identifier: oai:oa.upm.es:63755
Deposited by: Biblioteca Facultad de Informatica
Deposited on: 11 Sep 2020 06:12
Last Modified: 11 Sep 2020 06:12
  • Logo InvestigaM (UPM)
  • Logo GEOUP4
  • Logo Open Access
  • Open Access
  • Logo Sherpa/Romeo
    Check whether the anglo-saxon journal in which you have published an article allows you to also publish it under open access.
  • Logo Dulcinea
    Check whether the spanish journal in which you have published an article allows you to also publish it under open access.
  • Logo de Recolecta
  • Logo del Observatorio I+D+i UPM
  • Logo de OpenCourseWare UPM