• JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
 
  Bookmark and Share
 
 
Master's Dissertation
DOI
https://doi.org/10.11606/D.3.2005.tde-10112023-095922
Document
Author
Full name
Marcelo Cesar Cirelo
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2005
Supervisor
Committee
Cozman, Fabio Gagliardi (President)
Barros, Leliane Nunes de
Costa, Anna Helena Reali
Title in Portuguese
Uma metodologia para o aprendizado semi-supervisionado de classificadores Bayesianos.
Keywords in Portuguese
Aprendizado computacional
Inteligência artificial
Abstract in Portuguese
Neste trabalho são apresentados métodos para aprendizado de classificadores Bayesianos a partir de bases de dados contendo dados rotulados e não-rotulados (aprendizado semi-supervisionado). O trabalho apresenta dois novos algoritmos, SSS e CBL-EM, e compara estes algoritmos com versões de classificadores Naive Bayes, Tree-Augmented Naive Bayes e Structural-EM. As principais contribuições foram o desenvolvimento de um método para utilizar o algoritmo CBL1 em conjunto com o algoritmo EM (do inglês Expectation-Maximization) e a definição de uma metodologia para o aprendizado semi-supervisionado de classificadores Bayesianos. Os resultados empíricos mostram que os algoritmos propostos tem desempenho superior aos algoritmos existentes para aprendizado com dados rotulados e não-rotulados.
Title in English
Untitled in english
Keywords in English
Artificial intelligence
Computational learning
Abstract in English
This work presents techniques for learning Bayesian classifiers from databases containing labeled and unlabeled data (semi-supervised learning). The work presents two new algorithms, SSS and CBL-EM, and compares their performance with versions of Naive Bayes, Tree-Augmented Naive Bayes and Structural-EM classifiers. The main contributions of this work are the development of a framework for using the CBL1 and EM algorithms together, and the development of a methodology for the semi-supervised learning of Bayesian classifiers. The empirical tests show that the proposed algorithms perform better than existing classifiers for labeled and unlabeled data
 
WARNING - Viewing this document is conditioned on your acceptance of the following terms of use:
This document is only for private use for research and teaching activities. Reproduction for commercial use is forbidden. This rights cover the whole data about this document as well as its contents. Any uses or copies of this document in whole or in part must include the author's name.
Publishing Date
2023-11-10
 
WARNING: Learn what derived works are clicking here.
All rights of the thesis/dissertation are from the authors
CeTI-SC/STI
Digital Library of Theses and Dissertations of USP. Copyright © 2001-2024. All rights reserved.