• 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
 
 
Doctoral Thesis
DOI
https://doi.org/10.11606/T.3.2023.tde-23012024-101019
Document
Author
Full name
Orlando da Silva Junior
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2023
Supervisor
Committee
Almeida Junior, Jorge Rady de (President)
Oliveira, Rogério de
Prati, Ronaldo Cristiano
Sato, Liria Matsumoto
Zafalon, Geraldo Francisco Donegá
Title in Portuguese
Classificação de imagens em fluxos de dados utilizando aprendizado profundo.
Keywords in Portuguese
Aprendizado computacional
Aprendizagem profunda
Fluxos de dados
Processamento de imagens
Redes neurais
Abstract in Portuguese
As aplicações modernas tem transformado a vida em sociedade e trazido novos desafios computacionais. Um desses desafios é a produção massiva e não estacionária de dados de imagem, que requerem o desenvolvimento de algoritmos eficientes para processamento em tempo real e técnicas de análise para lidar com a natureza dinâmica desses dados. Nesta tese, a tarefa de classificação de imagens é explorada por meio de técnicas de aprendizagem em fluxos de dados. O objetivo é implementar algoritmos para a classificação de imagens utilizando aprendizado profundo, concentrando-se na investigação de modelos e algoritmos para o processamento e a aprendizagem de imagens em fluxos de dados. Um framework completo é proposto para a construção e validação de modelos das redes neurais profundas. Diferentes experimentos são conduzidos para avaliar a capacidade preditiva do framework em classificar novas imagens e os resultados são comparados com outros métodos do estado-da-arte. Os resultados mostram que o framework proposto supera os métodos comparados na maior parte dos cenários avaliados.
Title in English
Untitled in english
Keywords in English
Data streams
Deep learning
Image processing
Machine learning
Neural networks
Abstract in English
Modern applications have transformed life in society and brought new computational challenges. One of these challenges is the massive and non-stationary production of image data, which requires the development of efficient algorithms for real-time processing and analysis techniques to deal with the dynamic nature of these data. In this thesis, the task of image classification is explored through data stream learning techniques. The aim is to implement algorithms for image classification using deep learning, focusing on investigating models and algorithms for processing and learning images in data streams. A complete framework is proposed for the construction and validation of deep neural network models. Different experiments are conducted to evaluate the predictive capacity of the framework in classifying new images and the results are compared with other state-of-the-art methods. The results show that the proposed framework outperforms the compared methods in most of the evaluated scenarios.
 
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
2024-01-24
 
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.