• 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.45.2023.tde-03042023-095110
Document
Author
Full name
Marcela Musetti
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2023
Supervisor
Committee
Esteves, Luís Gustavo (President)
Diniz, Marcio Alves
Salasar, Luis Ernesto Bueno
Title in Portuguese
FBST em problemas de likelihood-free
Keywords in Portuguese
Aprendizado de máquina
FBST
Likelihood-free
Abstract in Portuguese
Na inferência Bayesiana, problemas com a obtenção da distribuição a posteriori analiticamente, em problemas de \textit são frequentes. E mesmo quando a solução desse problema é feito de forma computacional, ainda temos desafios com a realização de inferências, como o teste de hipóteses precisas. Uma forma Bayesiana de testar hipóteses precisas é via o Full Bayesian Significance Test (FBST) no qual calcula-se uma medida de evidência, denominada por e-valor. Nessa dissertação, queremos encontrar as densidades a posteriori em problemas de \textit pelo o método FlexCode, ao invés do tradicional MCMC e com as densidades em mãos propor uma solução para o cálculo do e-valor: através de métodos de classificação ou da integração computacional da densidade obtida via FlexCode. Dessa forma, conseguimos resolver problemas frequentes na inferência bayesiana transformando-os em um problema possível de resolver no universo de aprendizado de máquina
Title in English
Problems of likelihood-free by the FlexCode method
Keywords in English
FBST
Likelihood-free
Machine learning
Abstract in English
In Bayesian inference, issues with obtaining the analytics posterior distribution in likehood-free s problems are frequently. And yet, when the solution for these is made in a computational way, the challenges with the realization of inferences remains, like the precise hypothesis testing. One path to assemble the precise hypothesis testing is by the Full Bayesian Significance Test (FBST), calculating an evidence measure, the e-value. In this research, we aiming to find the posterior densities in problems of likehood-free by the FlexCode method, instead the traditional MCMC. Then, with the densities in hands, offer a solution to the e-value calculation: through the classification methods or the computational integration of the density obtained by FlexCode. Therefore, we can solve frequent problems in Bayesian inference by transforming then in a problem that could be solve in the machine learning universes.
 
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
2023-04-06
 
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.