Evolutionary Algorithms are search algorithms based on the Darwinian metaphor of “Natural Selection”. Typically these algorithms maintain a population of individual solutions, each of which has a fitness attached to it, which in some way reflects the quality of the solution. The search proceeds via the iterative generation, evaluation and possible incorporation of new individuals based on the current population, using a number of parameterised genetic operators. In this thesis the phenomenon of Self Adaptation of the genetic operators is investigated. A new framework for classifying adaptive algorithms is proposed, based on the scope of the adaptation, and on the nature of the transition function guiding the search through the space of possible configurations of the algorithm. Mechanisms are investigated for achieving the self adaptation of recombination and mutation operators within a genetic algorithm, and means of combining them are investigated. These are shown to produce significantly better results than any of the combinations of fixed operators tested, across a range of problem types. These new operators reduce the need for the designer of an algorithm to select appropriate choices of operators and parameters, thus aiding the implementation of genetic algorithms. The nature of the evolving search strategies are investigated and explained in terms of the known properties of the landscapes used, and it is suggested how observations of evolving strategies on unknown landscapes may be used to categorise them, and guide further changes in other facets of the genetic algorithm. This work provides a contribution towards the study of adaptation in Evolutionary Algorithms, and towards the design of robust search algorithms for “real world” problems.

Many industrial problems in various fields, such as logistics, process management, orproduct design, can be formalized and expressed as optimization problems in order tomake them solvable by optimization algorithms. However, solvers that guarantee thefinding of optimal solutions (complete) can in practice be unacceptably slow. Thisis one of the reasons why approximative (incomplete) algorithms, producing near-optimal solutions under restrictions (most dominant time), are of vital importance. Those approximative algorithms go under the umbrella term metaheuristics, each of which is more or less suitable for particular optimization problems. These algorithmsare flexible solvers that only require a representation for solutions and an evaluation function when searching the solution space for optimality.What all metaheuristics have in common is that their search is guided by certain control parameters. These parameters have to be manually set by the user andare generally problem and interdependent: A setting producing near-optimal resultsfor one problem is likely to perform worse for another. Automating the parameter setting process in a sophisticated, computationally cheap, and statistically reliable way is challenging and a significant amount of attention in the artificial intelligence and operational research communities. This activity has not yet produced any major breakthroughs concerning the utilization of problem instance knowledge or the employment of dynamic algorithm configuration. The thesis promotes automated parameter optimization with reference to the inverse impact of problem instance diversity on the quality of parameter settings with respect to instance-algorithm pairs. It further emphasizes the similarities between static and dynamic algorithm configuration and related problems in order to show how they relate to each other. It further proposes two frameworks for instance-based algorithm configuration and evaluates the experimental results. The first is a recommender system for static configurations, combining experimental design and machine learning. The second framework can be used for static or dynamic configuration,taking advantage of the iterative nature of population-based algorithms, which is a very important sub-class of metaheuristics. A straightforward implementation of framework one did not result in the expected improvements, supposedly because of pre-stabilization issues. The second approach shows competitive results in the scenario when compared to a state-of-the-art model-free configurator, reducing the training time by in excess of two orders of magnitude.

This thesis considers the problem of robust geometric regulation for tracking and disturbance rejection of systems governed by delay differential equations. It is well known that geometric regulation can be highly sensitive to system parameters and hence such designs are not always robust. In particular, when employing numerical approximations to delay systems, the resulting finite dimensional models inherit natural approximation errors that can impact robustness. This demonstrates this lack of robustness and then addresses robustness by employing versions of robust regulation that have been developed for infinite dimensional systems. Numerical examples are given to illustrate the ideas and to test the robustness of the regulator. / M.S. / Recent years have seen a surge in the everyday application of complex mechanical and electrical systems. These systems can perform complex tasks; however, the increased complexity makes it harder to control them. An example of such a system is a semi-autonomous car designed to stay within a designated lane. One of the most commonly used approaches for controlling such systems is called output regulation. In the above example, the output regulator regulates the output of the car (position of the car) to follow the reference output (the road lane). Traditionally, the design of output regulators assumes complete knowledge of the system. However, it is impossible to derive equations that govern complex systems like a car. This thesis analyzes the robustness of output regulators in the presence of errors in the system. In particular, the focus is on analyzing output regulators implemented to delay-differential equations. These are differential equations where the rate of change of states at the current time depends on the states at previous times. Furthermore, this thesis addresses this problem by employing the robust versions of the output regulators.

Scale is a widely used notion in medical image analysis that evolved in the form of scale-space theory where the key idea is to represent and analyze an image at various resolutions. Recently, a notion of local morphometric scale referred to as "tensor scale" was introduced using an ellipsoidal model that yields a unified representation of structure size, orientation and anisotropy. In the previous work, tensor scale was described using a 2-D algorithmic approach and a precise analytic definition was missing. Also, with previous framework, 3-D application is not practical due to computational complexity. The overall aim of the Ph.D. research is to establish an analytic definition of tensor scale in n-dimensional (n-D) images, to develop an efficient computational solution for 2- and 3-D images and to investigate its role in various medical imaging applications including image interpolation, filtering, and segmentation. Firstly, an analytic definition of tensor scale for n-D images consisting of objects formed by pseudo-Riemannian partitioning manifolds has been formulated. Tensor scale captures contextual structural information which is useful in local structure-adaptive anisotropic parameter control and local structure description for object/image matching. Therefore, it is helpful in a wide range of medical imaging algorithms and applications. Secondly, an efficient computational solution of tensor scale for 2- and 3-D images has been developed. The algorithm has combined Euclidean distance transform and several novel differential geometric approaches. The accuracy of the algorithm has been verified on both geometric phantoms and real images compared to the theoretical results generated using brute-force method. Also, a matrix representation has been derived facilitating several operations including tensor field smoothing to capture larger contextual knowledge. Thirdly, an inter-slice interpolation algorithm using 2-D tensor scale information of adjacent slices has been developed to determine the interpolation line at each image location in a gray level image. Experimental results have established the superiority of the tensor scale based interpolation method as compared to existing interpolation algorithms. Fourthly, an anisotropic diffusion filtering algorithm based on tensor scale has been developed. The method made use of tensor scale to design the conductance function for diffusion process so that along structure diffusion is encouraged and boundary sharpness is preserved. The performance has been tested on phantoms and medical images at various noise levels and the results were quantitatively compared with conventional gradient and structure tensor based algorithms. The experimental results formed are quite encouraging. Also, a tensor scale based n-linear interpolation method has been developed where the weights of neighbors were locally tuned based on local structure size and orientation. The method has been applied on several phantom and real images and the performance has been evaluated in comparison with standard linear interpolation and windowed Sinc interpolation methods. Experimental results have shown that the method helps to generate more precise structure boundaries without causing ringing artifacts. Finally, a new anisotropic constrained region growing method locally controlled by tensor scale has been developed for vessel segmentation that encourages axial region growing while arresting cross-structure leaking. The method has been successfully applied on several non-contrast pulmonary CT images. The accuracy of the new method has been evaluated using manually selection and the results found are very promising.

This master thesis focuses on the testing of the combustion equipment and the wood fuel combustion process. The aim of the thesis is to get acquainted with this issue and to analyze solutions for monitoring and control of heat source parameters according to valid legislation. The result of this thesis is the realization and verification of the software design of the combustion test system using the NI LabVIEW development environment.

The significant amount of data contained in digital images present barriers to methods of learning from the information they hold. Noise and the subjectivity of image evaluation further complicate such automated processes. In this thesis, we examine a particular area in which these difficulties are experienced. We attempt to control the parameters of a multi-step algorithm that processes visual information. A framework for approaching the parameter selection problem using reinforcement learning agents is presented as the main contribution of this research. We focus on the generation of state and action space, as well as task-dependent reward. We first discuss the automatic determination of fuzzy membership functions as a specific case of the above problem. Entropy of a fuzzy event is used as a reinforcement signal. Membership functions representing brightness have been automatically generated for several images. The results show that the reinforcement learning approach is superior to an existing simulated annealing-based approach. The framework has also been evaluated by optimizing ten parameters of the text detection for semantic indexing algorithm proposed by Wolf et al. Image features are defined and extracted to construct the state space. Generalization to reduce the state space is performed with the fuzzy ARTMAP neural network, offering much faster learning than in the previous tabular implementation, despite a much larger state and action space. Difficulties in using a continuous action space are overcome by employing the DIRECT method for global optimization without derivatives. The chosen parameters are evaluated using metrics of recall and precision, and are shown to be superior to the parameters previously recommended. We further discuss the interplay between intermediate and terminal reinforcement.

Neste trabalho foi realizada uma proposta de utilização de Sistemas de Inferência Nebulosos para controlar, em tempo de execução, parâmetros de Algoritmos Genéticos. Esta utilização busca melhorar o desempenho de Algoritmos Genéticos diminuindo, ao mesmo tempo: a média de iterações necessárias para que um Algoritmo Genético encontre o valor ótimo global procurado; bem como diminuindo o número de execuções do mesmo que não são capazes de encontrar o valor ótimo global procurado, nem mesmo para quantidades elevadas de iterações. Para isso, foram analisados os resultados de diversos experimentos com Algoritmos Genéticos, resolvendo instâncias dos problemas de Minimização de Funções e do Caixeiro Viajante, sob diferentes configurações de parâmetros. Com base nos resultados obtidos a partir destes experimentos, foi proposto um modelo com a troca de valores de parâmetros de Algoritmos Genéticos, em tempo de execução, pela utilização de Sistemas de Inferência Nebulosos, de forma a melhorar o desempenho do sistema, minimizando ambas as medidas citadas anteriormente. / This work addressed a proposal of the application of Fuzzy Systems to adjust parameters of Genetic Algorithms, during execution time. This application attempts to improve the performance of Genetic Algorithms by diminishing, at the same time: the average number of necessary generations for a Genetic Algorithm to find the desired global optimum value, as well as diminishing the number of executions of a Genetic Algorithm that are not capable of finding the desired global optimum value even for high numbers of generations. For that purpose, the results of many experiments with Genetic Algorithms were analyzed; addressing instances of the Function Minimization and the Travelling Salesman problems, under different parameter configurations. With the results obtained from these experiments, a model was proposed, for the exchange of parameter values of Genetic Algorithms, in execution time, by using Fuzzy Systems, in order to improve the performance of the system, minimizing both of the measures previously cited.

Neste trabalho foi realizada uma proposta de utilização de Sistemas de Inferência Nebulosos para controlar, em tempo de execução, parâmetros de Algoritmos Genéticos. Esta utilização busca melhorar o desempenho de Algoritmos Genéticos diminuindo, ao mesmo tempo: a média de iterações necessárias para que um Algoritmo Genético encontre o valor ótimo global procurado; bem como diminuindo o número de execuções do mesmo que não são capazes de encontrar o valor ótimo global procurado, nem mesmo para quantidades elevadas de iterações. Para isso, foram analisados os resultados de diversos experimentos com Algoritmos Genéticos, resolvendo instâncias dos problemas de Minimização de Funções e do Caixeiro Viajante, sob diferentes configurações de parâmetros. Com base nos resultados obtidos a partir destes experimentos, foi proposto um modelo com a troca de valores de parâmetros de Algoritmos Genéticos, em tempo de execução, pela utilização de Sistemas de Inferência Nebulosos, de forma a melhorar o desempenho do sistema, minimizando ambas as medidas citadas anteriormente. / This work addressed a proposal of the application of Fuzzy Systems to adjust parameters of Genetic Algorithms, during execution time. This application attempts to improve the performance of Genetic Algorithms by diminishing, at the same time: the average number of necessary generations for a Genetic Algorithm to find the desired global optimum value, as well as diminishing the number of executions of a Genetic Algorithm that are not capable of finding the desired global optimum value even for high numbers of generations. For that purpose, the results of many experiments with Genetic Algorithms were analyzed; addressing instances of the Function Minimization and the Travelling Salesman problems, under different parameter configurations. With the results obtained from these experiments, a model was proposed, for the exchange of parameter values of Genetic Algorithms, in execution time, by using Fuzzy Systems, in order to improve the performance of the system, minimizing both of the measures previously cited.

The significant amount of data contained in digital images present barriers to methods of learning from the information they hold. Noise and the subjectivity of image evaluation further complicate such automated processes. In this thesis, we examine a particular area in which these difficulties are experienced. We attempt to control the parameters of a multi-step algorithm that processes visual information. A framework for approaching the parameter selection problem using reinforcement learning agents is presented as the main contribution of this research. We focus on the generation of state and action space, as well as task-dependent reward. We first discuss the automatic determination of fuzzy membership functions as a specific case of the above problem. Entropy of a fuzzy event is used as a reinforcement signal. Membership functions representing brightness have been automatically generated for several images. The results show that the reinforcement learning approach is superior to an existing simulated annealing-based approach. The framework has also been evaluated by optimizing ten parameters of the text detection for semantic indexing algorithm proposed by Wolf et al. Image features are defined and extracted to construct the state space. Generalization to reduce the state space is performed with the fuzzy ARTMAP neural network, offering much faster learning than in the previous tabular implementation, despite a much larger state and action space. Difficulties in using a continuous action space are overcome by employing the DIRECT method for global optimization without derivatives. The chosen parameters are evaluated using metrics of recall and precision, and are shown to be superior to the parameters previously recommended. We further discuss the interplay between intermediate and terminal reinforcement.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2018-05-25T13:02:54Z No. of bitstreams: 1 guilhermedesouzaferreira.pdf: 1163831 bytes, checksum: ec0bec904b2e6110d9b9e4934727f35d (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2018-06-14T11:52:03Z (GMT) No. of bitstreams: 1 guilhermedesouzaferreira.pdf: 1163831 bytes, checksum: ec0bec904b2e6110d9b9e4934727f35d (MD5) / Made available in DSpace on 2018-06-14T11:52:03Z (GMT). No. of bitstreams: 1 guilhermedesouzaferreira.pdf: 1163831 bytes, checksum: ec0bec904b2e6110d9b9e4934727f35d (MD5) Previous issue date: 2018-02-22 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / O escalonamento de tarefas é um problema de otimização combinatória no qual tenta-se sequenciar da melhor maneira os trabalhos a serem realizados em processos de produção. O intuito neste caso é atingir os objetivos de desempenho estipulados pelo tomador de decisão, tais como, minimizar o makespan e minimizar o atraso total. O Problema de Sequencia-mento do tipo Job-Shop Flexível (FJSP) pertence a essa categoria, e caracteriza-se pela possibilidade de haver rotas tecnológicas diferentes para as tarefas e cada estágio poder ser composto por mais de uma máquina. Esse é o núcleo da tecnologia do gerenciamento de produção, pois sequenciamentos melhores podem encurtar o tempo de manufatura, reduzir os níveis de estoque, possibilitar a entrega de encomendas no tempo correto e aumentar a credibilidade dos processos e da empresa. Métodos exatos, que são computacionalmente custosos, são geralmente aplicados nos problemas de sequenciamento menores, portanto quando os problemas aumentam em tamanho, os métodos heurísticos e metaheurísticos começaram a ser aplicados. As metaheurísticas são importantes para solucionar FJSPs porque são mais rápidas do que os métodos exatos. Dentre elas, os Algoritmos Genéti-cos (AGs) estão entre as técnicas mais utilizadas para solucionar FJSPs e, atualmente, modelos híbridos vem sendo explorados, combinando AGs com técnicas de busca local e heurísticas para inicializar a população. No entanto, a escolha adequada dos parâmetros dos AGs é um trabalho difícil, recaindo num outro problema de otimização. Os Algoritmos Genéticos Adaptativos (AGAs) foram introduzidos para lidar com essa adversidade, uma vez que podem ajustar os parâmetros dos AGs durante o processo de busca. Portanto, o objetivo da presente dissertação é analisar diferentes técnicas adaptativas desenvolvidas para AGAs, com o intuito de reduzir o tempo de configuração dos AGs quando aplicados a FJSPs. Além disso, serão propostas alterações para as técnicas de atribuição de crédito e de seleção de operadores. Os estudos foram realizados em instâncias de diferentes tamanhos e os AGAs são comparados com AGs tradicionais. Duas diferentes análises foram realizadas baseadas em cenários no qual o tomador de decisão tem pouco tempo para configurar os algoritmos. Na Análise I, os AGAs tiveram desempenho semelhante aos AGs tradicionais, mas são interessantes por possuírem um menor número de parâmetros e, consequentemente, um menor tempo de configuração. Na Análise II, os AGAs geraram melhores resultados do que aqueles obtidos pelos AGs, o que os tornam apropriados para o caso em que há incerteza no processo produtivo e menor tempo de configuração. / Scheduling is a combinatorial optimization problem, in which one tries ordering the tasks to be performed in the processing units. The objective is to achieve the best values with respect to the performance indicators chosen by the decision-maker, such as, minimize the makespan and minimize the total lateness. The Flexible Job-Shop Scheduling Problem (FJSP) belongs to this category, and its characteristics are the different technological routes for the tasks and that each stage may consist of more than one machine. This is the technological core of the production management, as better schedules may reduce the manufacturing time, reduce the inventory, deliver the order in the right time, and raise the reliability of the process and the company. Exact methods, as they are computationally expensive, are usually employed for small scheduling problems, then heuristic and metaheuristic methods become interesting techniques for this type of problem. Metaheuristics are important to solve FJSPs as they are faster than the exact methods, and among then, Genetic Algorithms (GAs) are one of the most used techniques to solve FJSPs and, currently, they have been hybridized with local search and heuristics to initialize their population. However, to set up GAs is a hard-work and often generates another optimization problem. Adaptive Genetic Algorithms (AGAs) were introduced to work around this problem as they adapt the parameters of the GAs during the search process. Therefore, the objective of this dissertation is to analyze different adaptive techniques developed for AGAs with the purpose of reducing the setup time of GAs when they are applied to FJSPs. In addition, modifications will be proposed for the operator selection techniques and for credit assignment schemes. The studies were performed in instances of different sizes, and the AGAs are compared with traditional GAs. Two different analyzes were performed based on scenarios in which the decision maker does not has to much time to configure the algorithms. In Analysis I, some AGAs performed similarly to the traditional GAs, but they are more interesting as they have a smaller number of parameters, thus a shorter configuration time. In Analysis II, some AGAsgeneratedbetterresultsthanthoseobtainedbyGAs, whichmakesthemappropriate for the case when there is uncertainty in the production process and the decision maker does not have too much time to configure the algorithm.