The goal of this master thesis is to examine possibilities of realizing comptutational structures in cellular automata. The work describes the fundamental principles of cellular automata and summarizes some ways of how to achive the specified goal. An overview of Turing-complete and other specialized computational tasks is proposed considering both 1D and 2D cellular automata. It is shown that different computational scenarios in cellular automata can be considered with various setups of the input and output arrangements. With regard to showed inputs and outputs arrangement, sets of tests is designed to find solutions of choosen computational structures on cellular automata with use of choosen evolutionary algorithm. Found solutions are compared by computational resources consumption and difficulty of discovery later.

The goal of this Master's Thesis is finding a SAT solving heuristic by the application of an evolutionary algorithm. This thesis surveys various approaches used in SAT solving and some variants of evolutionary algorithms that are relevant to this topic. Afterwards the implementation of a linear genetic programming system that searches for a suitable heuristic for SAT problem instances is described, together with the implementation of a custom SAT solver which expoloits the output of the genetic program. Finally, the achieved results are summarized.

Numerical optimization of multimodal or otherwise nontrivial functions has stayed around the peak of the interest of many researchers for a long time. One of the promising methods that appeared is the hybrid approach of the Learnable Evolution Model that combines the well-established ways of artificial intelligence and machine learning with recently popular and efective methods of evolutionary programming. In this work, the method itself was reviewed with respect to what has been already implemented and tested and several possible new implementations of the method were proposed and some of them consequently implemented. The resulting program was then tested against a set of chosen nontrivial real-valued functions and its results were compared to those achieved with EDA algorithms.

Symbolic regression is the problem of identifying the mathematic description of a hidden system from experimental data. Symbolic regression is closely related to general machine learning. This work deals with symbolic regression and its solution based on the principle of genetic programming and coevolution. Genetic programming is the evolution based machine learning method, which automaticaly generates whole programs in the given programming language. Coevolution of fitness predictors is the optimalization method of the fitness modelling that reduces the fitness evaluation cost and frequency, while maintainig evolutionary progress. This work deals with concept and implementation of the solution of symbolic regression using coevolution of fitness predictors, and its comparison to a solution without coevolution. Experiments were performed using cartesian genetic programming.

This diploma thesis deals with acceleration of advanced genetic algorithm. For implementation, discrete and continuos versions of UMDA genetic algorithm were chosen. The main part of the acceleration is the utilization of SSE instruction set. Using this set, the functions for calculating fitness and new population sampling were accelerated in particular. Then the pseudorandom number generator that also uses SSE instruction set was implemented.  The discrete algorithm reached the speed of up to 4,6 after this implementation. Finally, the algorithms were modified so that the system  OpenMP could be used, which enables the running of blocks of code in more threads. The continuous version of algorithm is not convenient for parallelization, because computational complexity of that algorithm is low. In comparison, the discrete versions of algorithm are really appropriate for parallelization. Both the implemented versions reached the total acceleration of up to 4,9 and 7,2.

Resumo: O principal objetivo deste estudo é realizar uma análise de aspectos críticos que surgem na modelagem matemática do problema de planejamento da expansão de sistemas de transmissão a longo prazo, assim como o desenvolvimento de ferramentas computacionais para a prova de novos modelos e metodologias que possam contribuir na solução do problema de planejamento de sistemas de transmissão de energia elétrica considerando as condições dos sistemas modernos de energia elétrica. Com esta metodologia, busca-se obter uma rede de transmissão mais eficiente, e com o menor custo possível, que se adapte as novas exigências produzidas pela introdução da desregulação nos sistemas elétricos. Para isto combinam-se três aspectos: rede futura livre de congestionamento, desplanificação e incerteza na geração e na demanda futura, os quais são manipuladas desde a perspectiva mono-objetivo e multiobjetivo. A possibilidade de eliminar completamente o congestionamento na rede de transmissão é analisada através da inclusão no modelo de todos os cenários de geração factíveis futuros, e não somente alguns cenários como outros estudos. Considerar uma operação sem congestionamento para o futuro está associado a grandes custos de investimento. Para atenuar este grande custo uma opção é incluir a possibilidade de desplanificação e a inclusão dos efeitos das incertezas presentes na geração e na demanda futura no problema de planejamento. O problema de planejamento de sistemas de transmissão é um problema de programação não linear inteira mista (PNLIM) quando é usado o modelo DC. Praticamente todos os algoritmos usados para resolver este problema utilizam uma sub-rotina de programação linear (PL) para resolver problemas de PL resultantes do algoritmo de solucão do problema de planejamento, os quais são denominados... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: The main goal for this study is to do an analysis of the critical issues that appear in the mathematical modeling of the transmission system expansion planning problem, when long term is considered. A methodology was developed and a computational tool, to solve the transmission expansion planning in modern electrical systems. With this methodology more efficient electrical networks are obtained, at low investment costs. This is accomplished taking into account three important aspects: open access, or congestion-free planning, uncertainty in demand and generation, and de-planning. The problem is solved using mono-objective and multi-objective methodologies. For this investigation, congestion-free transmission networks should consider all the future and feasible scenarios of generation, unlike some papers, where only a few scenarios are taken in to account. This feature is associated to high investment costs. Lower costs are often obtained by the inclusion of uncertainty in future demand and future generation. The transmission system expansion planning problem is a no-linear integer-mixed programming problem (PNLIM) when the DC model is used. Practically, all the algorithms used in the solution process, for this problem, use one subroutine of linear programming (PL) for solved the PL problems that result during the solution process, in the denominated operative problem. The solution of the PL's is the part of the problem that requires the biggest computational effort, because during the solution process is necessary to solved thousands or millions of PL's, for high size problems. the PNLIM problem is solved through the combination of a meta-heuristic method and a linear programming method. The meta-heuristic method solves the denominated investment problem and the PL the denominated operational problem. The transmission planning problem considering multiples generation scenarios... (Complete abstract click electronic access below) / Orientador: Rubén Augusto Romero Lázaro / Coorientador: José Roberto Sanches Mantovani / Banca: Carlos Roberto Minussi / Banca: Sérgio Azevedo de Oliveira / Banca: Ariovaldo Verandio Garcia / Banca: Ramón Alfonso Gallego Rendón / Doutor

Dans le contexte du Web sémantique, les ontologies OWL représentent des connaissances explicites sur un domaine sur la base d'une conceptualisation des domaines d'intérêt, tandis que la connaissance correspondante sur les individus est donnée par les données RDF qui s'y réfèrent. Dans cette thèse, sur la base d'idées dérivées de l'ILP, nous visons à découvrir des motifs de connaissance cachés sous la forme de règles d'association multi-relationnelles en exploitant l'évidence provenant des assertions contenues dans les bases de connaissances ontologiques. Plus précisément, les règles découvertes sont codées en SWRL pour être facilement intégrées dans l'ontologie, enrichissant ainsi son pouvoir expressif et augmentant les connaissances sur les individus (assertions) qui en peuvent être dérivées. Deux algorithmes appliqués aux bases de connaissances ontologiques peuplées sont proposés pour trouver des règles à forte puissance inductive : (i) un algorithme de génération et test par niveaux et (ii) un algorithme évolutif. Nous avons effectué des expériences sur des ontologies accessibles au public, validant les performances de notre approche et les comparant avec les principaux systèmes de l'état de l'art. En outre, nous effectuons une comparaison des métriques asymétriques les plus répandues, proposées à l'origine pour la notation de règles d'association, comme éléments constitutifs d'une fonction de fitness pour l'algorithme évolutif afin de sélectionner les métriques qui conviennent à la sémantique des données. Afin d'améliorer les performances du système, nous avons proposé de construire un algorithme pour calculer les métriques au lieu d'interroger viaSPARQL-DL. / In the Semantic Web context, OWL ontologies represent explicit domain knowledge based on the conceptualization of domains of interest while the corresponding assertional knowledge is given by RDF data referring to them. In this thesis, based on ideas derived from ILP, we aim at discovering hidden knowledge patterns in the form of multi-relational association rules by exploiting the evidence coming from the assertional data of ontological knowledge bases. Specifically, discovered rules are coded in SWRL to be easily integrated within the ontology, thus enriching its expressive power and augmenting the assertional knowledge that can be derived. Two algorithms applied to populated ontological knowledge bases are proposed for finding rules with a high inductive power: (i) level-wise generated-and-test algorithm and (ii) evolutionary algorithm. We performed experiments on publicly available ontologies, validating the performances of our approach and comparing them with the main state-of-the-art systems. In addition, we carry out a comparison of popular asymmetric metrics, originally proposed for scoring association rules, as building blocks for a fitness function for evolutionary algorithm to select metrics that are suitable with data semantics. In order to improve the system performance, we proposed to build an algorithm to compute metrics instead of querying via SPARQL-DL.

Despite the fact that cogeneration (CHP) and tri-generation (CHCP) are among the most efficient ways to produce electricity and thermal energy, there is still some unexploited potential for these techniques. One could say that the circumstances for using these techniques are better now than ever. Some of the reasons for applying CHP and CHCP are: the techniques are well understood, their application could generate some profit, and the required technology is available. Moreover, there is increasing concern in regards to energy security, the need to increase the energy efficiency in power generation and distribution as well as to lower the emissions from fossil fuel combustion. CHP/CHCP promoters and developers face difficulties when analyzing the conditions and proposing a plan of application. On one hand, there are some external barriers which have to be torn down by means of energy regulation schemes. These may include economic incentives, easy and safe interconnection to the grid to export electricity and have backup if necessary, and access to the market to sell the surplus of electricity at a fair price. On the other hand, there are some internal barriers such as the difficulty evaluating potential energy savings, emission reduction, and economic performance of a project based on the circumstances of a specific site; lack of awareness; unwillingness to invest in CHP/CHCP projects; and difficulty in selecting and sizing the equipment which would give the maximum benefits in terms of life cycle cost, energy savings and emission reduction. Nowadays, it is possible to develop software tools which use simulations and optimization algorithms to evaluate several options, compare them and chose the ones that give the optimum performance with respect to an objective function defined by the user. In this project, the general context for the application of cogeneration and tri-generation projects was studied including factors which have an impact on its feasibility and performance. Moreover, a survey of the exiting feasibility analysis tools was done, and a case study was chosen and analyzed. Next, a model was developed using the software Trnsys for the simulation and Matlab for the optimization. The model was tested by evaluating the study case. The result of the simulation and optimization gives several possible equipment size combinations. The tradeoff between two different objective functions such as net present value and primary energy savings or emission reduction is presented in Pareto front diagrams. The main conclusion of this project is that by using Trnsys and Matlab, it is possible to develop more complex models which, when applying optimization algorisms, could become a very useful and helpful tool that CHP/CHCP developers could use to speed up the analysis of projects while contributing to the goal of deploying these techniques.

Multi-objective evolutionary algorithms have gained a lot of atten- tion in the recent years. They have proven to be among the best multi-objective optimizers and have been used in many industrial ap- plications. However, their usability is hindered by the large number of evaluations of the objective functions they require. These can be expensive when solving practical tasks. In order to reduce the num- ber of objective function evaluations, surrogate models can be used. These are a simple and fast approximations of the real objectives. In this work we present the results of research made between the years 2009 and 2013. We present a multi-objective evolutionary algo- rithm with aggregate surrogate model, its newer version, which also uses a surrogate model for the pre-selection of individuals. In the next part we discuss the problem of selection of a particular type of model. We show which characteristics of the various models are im- portant and desirable and provide a framework which combines sur- rogate modeling with meta-learning. Finally, in the last part, we ap- ply multi-objective optimization to the problem of hyper-parameters tuning. We show that additional objectives can make finding of good parameters for classifiers faster. 1

[pt] O envelhecimento da voz, conhecido como presbifonia, é um processo natural que pode causar grande modificação na qualidade vocal do indivíduo. A sua identificação precoce pode trazer benefícios, buscando tratamentos que possam prevenir o seu avanço. Esse trabalho tem como motivação a identificação de vozes com sinais de envelhecimento através de redes neurais do tipo Spiking (SNN). O objetivo principal é o de construir dois novos modelos, denominados híbridos, utilizando SNN para problemas de agrupamento, onde os atributos de entrada e os parâmetros que configuram a SNN são otimizados por algoritmos evolutivos. Mais especificamente, os modelos neuro-evolucionários propostos são utilizados com o propósito de configurar corretamente a SNN, e selecionar os atributos mais relevantes para a formação dos grupos. Os algoritmos evolutivos utilizados foram o Algoritmo Evolutivo com Inspiração Quântica com representação Binário-Real (AEIQ-BR) e o Optimization by Genetic Programming (OGP). Os modelos resultantes foram nomeados Quantum-Inspired Evolution of Spiking Neural Networks with Binary-Real (QbrSNN) e Spiking Neural Network Optimization by Genetic Programming (SNN-OGP). Foram utilizadas oito bases benchmark e duas bases de voz, masculinas e femininas, a fim de caracterizar o envelhecimento. Para uma análise funcional da SNN, as bases benchmark forma testadas com uma abordagem clássica de agrupamento (kmeans) e com uma SNN sem evolução. Os modelos propostos foram comparados com uma abordagem clássica de Algoritmo Genético (AG). Os resultados mostraram a viabilidade do uso das SNNs para agrupamento de vozes envelhecidas. / [en] The aging of the voice, known as presbyphonia, is a natural process that can cause great change in vocal quality of the individual. Its early identification can benefit, seeking treatments that could prevent their advance. This work is motivated by the identification of voices with signs of aging through neural networks of spiking type (SNN). The main objective is to build two new models, called hybrids, using SNN for clustering problems where the input attributes and parameters that configure the SNN are optimized by evolutionary algorithms. More specifically, the proposed neuro-evolutionary models are used in order to properly configure the SNN, and select the most relevant attributes for the formation of groups. Evolutionary algorithms used were the Evolutionary Algorithm with Quantum Inspiration with representation Binary-Real (AEIQ-BR) and the Optimization by Genetic Programming (OGP). The resulting models were named Quantum-Inspired Spiking Neural Evolution of Networks with Binary-Real (QbrSNN) and Spiking Neural Network Optimization by Genetic Programming (SNN-OGP). Eight bases were used, and two voice benchmark bases, male and female, in order to characterize aging. NNS for functional analysis, the tested benchmark base form with a classical clustering approach (kmeans) and a SNN without change. The proposed models were compared with a classical approach of Genetic Algorithm (GA). The results showed the feasibility of using the SNN to agrupamentode aged voices.