• 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.45.2012.tde-11092012-134255
Document
Author
Full name
Rosario Alejandra Medina Rodríguez
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2012
Supervisor
Committee
Hashimoto, Ronaldo Fumio (President)
Hirata Junior, Roberto
Reali Costa, Anna Helena
Title in Portuguese
Algoritmos Evolutivos aplicados ao Classificador baseado em Segmentos de Reta
Keywords in Portuguese
algoritmos de otimizacao
algoritmos evo- lutivos
aprendizado computacional supervisionado
reconhecimento de padroes
Segmentos de reta
Abstract in Portuguese
Nos ultimos anos o uso de tecnicas de aprendizado computacional tornou se uma das tarefas comumente realizadas, pois tem inumeras aplicacoes de reconhecimento de padroes, tais como: reco- nhecimento de voz, classificacao de texto, reconhecimento facial, diagnostico por imagens medicas, entre outras. Dessa forma, um grande numero de tecnicas que lidam com este tipo de problema tem sido desenvolvido ate o momento. Neste trabalho apresentamos uma alternativa para melhorar a taxa acerto de classificacao do classificador binario SLS, que apresentou resultados comparaveis com as SVMs. Nesse metodo, o Gradiente Descendente e utilizado para otimizar a posicao final dos conjuntos de segmentos de reta que representarao cada classe. Embora convirja rapidamente a um valor otimo, muitas vezes e possivel o algoritmo parar em uma regiao de otimos locais, que nao representa o minimo global. Dado esse problema, foram utilizados diferentes algoritmos evolutivos em combinacao com o Gradiente Descendente a fim de melhorar a acuracia do classificador SLS. Adicionalmente a aplicacao de algoritmos evolutivos na fase de treinamento do classificador SLS, foram exploradas duas propostas: (i) explorar o uso de diferente numero de segmentos de reta para representar a distribuicao de dados de cada classe. Dado que no algoritmo original do metodo SLS o numero de segmentos de reta e igual para cada classe, o qual pode significar alguma perda de acuracia ou sobreposicao dos segmentos de reta; (ii) estimar a melhor combinacao de segmentos de reta a serem usados para cada classe. O uso de diferentes quantidades de segmentos de reta por classe pode ser de ajuda na obtencao de melhores porcentagens de acerto, mas determinar uma quantidade otima que permita representar cada classe, e um trabalho dificil. Assim, usamos o algoritmo X-Means, que e um algoritmo de agrupamento, para estimar o numero de segmentos de reta. As propostas exibiram bons resultados que possibilitam a aplicacao do classificador SLS, com um algoritmo de treinamento hibrido, em problemas reais.
Title in English
Evolutive Algorithms applied to the Straight Line Segment Classifier
Keywords in English
evolutive algorithms
optimization algorithms
pattern recognition.
straight line segments
supervised machine learning
Abstract in English
During the past years, the use of machine learning techniques have become into one of the most frequently performed tasks, due to the large amount of pattern recognition applications such as: voice recognition, text classification, face recognition, medical image diagnosis, among others. Thus, a great number of techniques dealing with this kind of problem have been developed until now. In this work, we propose an alternative training algorithm to improve the accuracy of the SLS binary Classifier, which produces good results that can be compared to Support Vector Machines. In that classifier, the Gradient Descent method has been used to optimize the final positions of two sets of straight line segments that represent each class. Although, this method quickly converges to an optimum, it is possible that the algorithm stops at a local optimum region, which does not guarantee a global minimum. Given that problem, we combine evolutive optimization algorithms with the gradient descent method to improve the accuracy of the SLS Classifier. In addition to our proposal of using evolutive algorithms, we also developed two proposals: (i) we explore the use of different number of straight line segments to represent the data distribution. Since the original SLS classifier algorithm uses the same number of segments for each class, which could lead to a loss of accuracy or straight line segments overlapping. So, using different number of segments could be the way to improve the accuracy; (ii) estimate the best combination of straight line segments to represent each class. Finding an optimal combination, can be a very difficult problem, so we propose the X-Means algorithm to determine the number of segments. The proposed methodology showed good results which can be used to solve some other real problems with the SLS classifier using the proposed hybrid training algorithm.
 
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
2012-09-13
 
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.