• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 1
  • 1
  • 1
  • Tagged with
  • 3
  • 3
  • 3
  • 3
  • 3
  • 2
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

The Problem of Tuning Metaheuristics as seen from a Machine Learning Perspective

Birattari, Mauro 20 December 2004 (has links)
<p>A metaheuristic is a generic algorithmic template that, once properly instantiated, can be used for finding high quality solutions of combinatorial optimization problems. For obtaining a fully functioning algorithm, a metaheuristic needs to be configured: typically some modules need to be instantiated and some parameters need to be tuned. For the sake of precision, we use the expression <em>parametric tuning</em> for referring to the tuning of numerical parameters, either continuous or discrete but in any case ordinal. On the other hand, we use the expression <em>structural tuning</em> for referring to the problem of defining which modules should be included and, in general, to the problem of tuning parameters that are either boolean or categorical. Finally, with <em>tuning</em> we refer to the composite <em>structural and parametric tuning</em>.</p> <p>Tuning metaheuristics is a very sensitive issue both in practical applications and in academic studies. Nevertheless, a precise definition of the tuning problem is missing in the literature. In this thesis, we argue that the problem of tuning a metaheuristic can be profitably described and solved as a machine learning problem.</p> <p>Indeed, looking at the problem of tuning metaheuristics from a machine learning perspective, we are in the position of giving a formal statement of the tuning problem and to propose an algorithm, called F-Race, for tackling the problem itself. Moreover, always from this standpoint, we are able to highlight and discuss some catches and faults in the current research methodology in the metaheuristics field, and to propose some guidelines.</p> <p>The thesis contains experimental results on the use of F-Race and some examples of practical applications. Among others, we present a feasibility study carried out by the German-based software company <em>SAP</em>, that concerned the possible use of F-Race for tuning a commercial computer program for vehicle routing and scheduling problems. Moreover, we discuss the successful use of F-Race for tuning the best performing algorithm submitted to the <em>International Timetabling Competition</em> organized in 2003 by the <em>Metaheuristics Network</em> and sponsored by <em>PATAT</em>, the international series of conferences on the <em>Practice and Theory of Automated Timetabling</em>.</p>
2

Uma heurística para otimização de meta-heurísticas por meio de métodos estatísticos / A heuristic for optimization of metaheuristics by means of statistical methods

Barbosa, Eduardo Batista de Moraes [UNESP] 01 July 2016 (has links)
Submitted by EDUARDO BATISTA DE MORAES BARBOSA null (ebmb@yahoo.com) on 2016-07-22T20:43:38Z No. of bitstreams: 1 Thesis-Full.pdf: 4249671 bytes, checksum: 293e98d71cda47dab135797fedb06e6f (MD5) / Approved for entry into archive by Ana Paula Grisoto (grisotoana@reitoria.unesp.br) on 2016-07-25T17:18:40Z (GMT) No. of bitstreams: 1 barbosa_ebm_dr_guara.pdf: 4249671 bytes, checksum: 293e98d71cda47dab135797fedb06e6f (MD5) / Made available in DSpace on 2016-07-25T17:18:40Z (GMT). No. of bitstreams: 1 barbosa_ebm_dr_guara.pdf: 4249671 bytes, checksum: 293e98d71cda47dab135797fedb06e6f (MD5) Previous issue date: 2016-07-01 / A configuração de parâmetros de algoritmos, em especial, das meta-heurísticas, nem sempre é trivial e, frequentemente, é realizada ad hoc de acordo com o problema sob análise. A fim de resolver o problema de sintonização de meta-heurísticas, a presente pesquisa propõe uma metodologia que combina o uso de técnicas estatísticas robustas (ex.: Planejamento de Experimentos) e métodos eficientes de Inteligência Artificial (ex.: Algoritmos de Corrida). A ideia central desta metodologia é um método heurístico, denominado Algoritmo de Corrida Orientada por Heurística (HORA), capaz de explorar o espaço de busca para perseguir diferentes alternativas na vizinhança de uma configuração de parâmetros promissora e encontrar sistematicamente boas configurações candidatas para diferentes algoritmos. Em síntese, o método HORA concentra as buscas sobre configurações candidatas promissoras, criadas dinamicamente em um processo iterativo, e utiliza uma técnica estatística robusta para avaliar as diferentes alternativas e descartar aquelas de qualidade inferior, assim que reunir evidências estatísticas suficientes contra elas. A partir dos resultados de diversos estudos computacionais, em que diferentes meta-heurísticas foram aplicadas sobre dois problemas clássicos de otimização combinatória, apresentam-se evidências estatísticas que as sintonizações obtidas pelo HORA são competitivas em relação ao método de Corrida e seu tempo no processo de sintonização é amplamente vantajoso. Em um estudo complementar, um algoritmo já bem configurado da literatura foi sintonizado por meio da metodologia proposta e os resultados da nova sintonização foram comparados com a literatura. Os resultados demonstram que a sintonização obtida pelo HORA pode encontrar soluções de melhor qualidade em relação à sintonização original. Portanto, a partir dos resultados apresentados nesta pesquisa conclui-se que a metodologia para sintonização de meta-heurísticas por meio do método HORA é uma abordagem promissora que pode ser aplicada sobre diferentes meta-heurísticas para resolução de uma diversidade de problemas de otimização. / The fine-tuning of the algorithms parameters, specially, of the meta-heuristics, it is not always trivial and often is performed by ad hoc methods according to the problem under analysis. In order to solve the problem of tuning metaheuristics, this research proposes a methodology combining statistical robust techniques (e.g.: Design of Experiments) and efficient methods from Artificial Intelligence (e.g.: Racing Algorithms). The key idea of this methodology is a heuristic method, called Heuristic Oriented Racing Algorithm (HORA), which explores the search space looking for alternatives near of a promising candidate and consistently finds good candidates configuration for different algorithms. Briefly, HORA focuses its searches over the promising candidates configuration, dynamically created in an iterative process, and employs a robust statistical method to evaluate and discarding them, as soon as gather enough statistical evidence against them. The results of several studies, where different metaheuristics were applied to solve two classical combinatorial optimization problems, present statistical evidences that the settings obtained by HORA are competitive to the Racing Algorithms and its time in the fine-tuning process is widely advantageous. In a complementary study, an already well setting algorithm from the literature was tuned by means of the proposed methodology and the new settings were compared with the literature. The results show that the fine-tuning from HORA can find better quality solutions than the original ones. Therefore, from the results presented in this study it is concluded that the methodology for fine-tuning of metaheuristics by means of HORA is a promising approach, which can be applied on different metaheuristics to solve a diversity of optimization problems.
3

The problem of tuning metaheuristics as seen from a machine learning perspective

Birattari, Mauro 20 December 2004 (has links)
<p>A metaheuristic is a generic algorithmic template that, once properly instantiated, can be used for finding high quality solutions of combinatorial optimization problems.<p>For obtaining a fully functioning algorithm, a metaheuristic needs to be configured: typically some modules need to be instantiated and some parameters need to be tuned. For the sake of precision, we use the expression <em>parametric tuning</em> for referring to the tuning of numerical parameters, either continuous or discrete but in any case ordinal. <p>On the other hand, we use the expression <em>structural tuning</em> for referring to the problem of defining which modules should be included and, in general, to the problem of tuning parameters that are either boolean or categorical. Finally, with <em>tuning</em> we refer to the composite <em>structural and parametric tuning</em>.</p><p><p><p>Tuning metaheuristics is a very sensitive issue both in practical applications and in academic studies. Nevertheless, a precise definition of the tuning problem is missing in the literature. In this thesis, we argue that the problem of tuning a metaheuristic can be profitably described and solved as a machine learning problem.</p><p><p><p>Indeed, looking at the problem of tuning metaheuristics from a machine learning perspective, we are in the position of giving a formal statement of the tuning problem and to propose an algorithm, called F-Race, for tackling the problem itself. Moreover, always from this standpoint, we are able to highlight and discuss some catches and faults in the current research methodology in the metaheuristics field, and to propose some guidelines.</p><p><p><p>The thesis contains experimental results on the use of F-Race and some examples of practical applications. Among others, we present a feasibility study carried out by the German-based software company <em>SAP</em>, that concerned the possible use of F-Race for tuning a commercial computer program for vehicle routing and scheduling problems. Moreover, we discuss the successful use of F-Race for tuning the best performing algorithm submitted to the <em>International Timetabling Competition</em> organized in 2003 by the <em>Metaheuristics Network</em> and sponsored by <em>PATAT</em>, the international series of conferences on the <em>Practice and Theory of Automated Timetabling</em>.</p> / Doctorat en sciences appliquées / info:eu-repo/semantics/nonPublished

Page generated in 0.0838 seconds