Des travaux récents ont montré l’intérêt de la programmation par contraintes pour la fouille de données. Dans cette thèse, nous nous intéressons à la recherche de motifs sur séquences, et en particulier à la caractérisation, à l’aide de motifs, de classes de séquences pré-établies. Nous proposons à cet effet un langage de modélisation à base de contraintes qui suppose une représentation matricielle du jeu de séquences. Un motif s’y définit comme un ensemble de caractères (ou de patrons) et pour chacun une localisation dans différentes séquences. Diverses contraintes peuvent alors s’appliquer : validité des localisations, couverture d’une classe de séquences, ordre sur les localisations des caractères commun aux séquences, etc. Nous formulons deux problèmes de caractérisation NP-complets : la caractérisation par motif totalement ordonné (e.g. sous-séquence exclusive à une classe) ou partiellement ordonné. Nous en donnons deux modélisations CSP qui intègrent des contraintes globales pour la preuve d’exclusivité. Nous introduisons ensuite un algorithme mémétique pour l’extraction de motifs partiellement ordonnés qui s’appuie sur la résolution CSP lors des phases d’initialisation et d’intensification. Cette approche hybride se révèle plus performante que l’approche CSP pure sur des séquences biologiques. La mise en forme matricielle de jeux de séquences basée sur une localisation des caractères peut être de taille rédhibitoire. Nous proposons donc de localiser des patrons plutôt que des caractères. Nous présentons deux méthodes ad-hoc, l’une basée sur un parcours de treillis et l’autre sur la programmation dynamique. / Recent works have shown the relevance of constraint programming to tackle data mining tasks. This thesis follows this approach and addresses motif discovery in sequential data. We focus in particular, in the case of classified sequences, on the search for motifs that best fit each individual class. We propose a language of constraints over matrix domains to model such problems. The language assumes a preprocessing of the data set (e.g., by pre-computing the locations of each character in each sequence) and views a motif as the choice of a sub-matrix (i.e., characters, sequences, and locations). We introduce different matrix constraints (compatibility of locations with the database, class covering, location-based character ordering common to sequences, etc.) and address two NP-complete problems: the search for class-specific totally ordered motifs (e.g., exclusive subsequences) or partially ordered motifs. We provide two CSP models that rely on global constraints to prove exclusivity. We then present a memetic algorithm that uses this CSP model during initialisation and intensification. This hybrid approach proves competitive compared to the pure CSP approach as shown by experiments carried out on protein sequences. Lastly, we investigate data set preprocessing based on patterns rather than characters, in order to reduce the size of the resulting matrix domain. To this end, we present and compare two alternative methods, one based on lattice search, the other on dynamic programming.

De nombreux problèmes de décision issus du monde réel sont de nature NP-difficile. Il est également fréquent que de tels problèmes rassemblent plusieurs objectifs à optimiser simultanément, généralement contradictoires entre eux. Pour aborder cette classe de problèmes, les métaheuristiques multiobjectifs fournissent des outils particulièrement efficaces. Par ailleurs, pour traiter des problèmes de transport, l'élaboration de modèles permettant de caractériser l’évolution spatio-temporelle d’une population est un élément essentiel. Dans le cadre de ces travaux, nous nous intéressons à la chaine complète qui permet de guider une décision dans le domaine de l'aménagement du territoire et du transport. Nous considérons ainsi les deux principales phases impliquées dans le processus de décision : la modélisation des déplacements de la population d'une part, et l'élaboration d'une métaheuristique hybride pour résoudre des problèmes d'optimisation multiobjectif d'autre part. Afin de modéliser l’évolution de la présence de personnes sur un territoire, nous proposons dans cette thèse un nouveau modèle de mobilité. L'originalité de ce travail réside dans l'utilisation de données nouvelles issues de la téléphonie mobile, ainsi que dans l'exploitation d'informations géographiques et socio-économiques pour caractériser le pouvoir d'attraction du territoire. Nous proposons par ailleurs une heuristique pour résoudre des problèmes multiobjectifs. L’étude de l'influence de différents opérateurs sur la construction de l'ensemble Pareto, nous a amené à concevoir une heuristique hybride de type mémétique, qui se révèle être significativement plus efficace que des approches de référence. Les deux principales phases, modélisation et optimisation, ont été expérimentées et validées dans un contexte réel. Elles ont donné lieu au développement d’une plate-forme logicielle d’aide à la décision utilisée notamment pour proposer des emplacements de stations pour un service d'auto-partage électrique. / Many decision-making problems in the real world are NP-hard. These problems commonly feature several mutually-contradictory objectives to be optimized simultaneously. Multiobjective metaheuristics provide particularly effective means of addressing this class of problems. Moreover, for transportation problems, the development of models able to evaluate the spatiotemporal evolution of a population is essential. In our research, we are interested in the complete chain guiding a decision in the fields of transportation and territory planning. We consider the two main phases involved in the decision-making process: building a population mobility model and developing a hybrid metaheuristic to solve multiobjective optimization problems. In order to compute the evolution of population presence on a territory, in this thesis we propose a new mobility model; its originality lies in employing new data from mobile phone networks as well as geographic and socio-economic information to indicate the attractiveness of the territory. We have also developed a heuristic to solve multiobjective problems: following the study of the influence of several operators on the Pareto front, we have designed a hybrid memetic heuristic that is significantly more effective than reference approaches. The two main phases of modelling and optimizing have been tested and validated in a real context, allowing us to develop a decision-making software platform that can be used to provide station locations for an electric car-sharing service.

This thesis deals with an optimization of waste collection in a mid-sized town. The model is formulated based on requirements from a real process. To deal with this problem, the original memetic algorithm was developed and implemented in C++.

This thesis deals with the problem of vehicle scheduling in public transport. It contains a theoretical introduction to vehicles scheduling and evolutionary algorithms. Vehicle scheduling is analyzed with respect to the bus timetables. Analysis of evolutionary algorithms is done with emphasis on the genetic algorithms and tabu-search method After the theoretical introduction, a memetic algorithm for the given problem is analyzed. Finally, the thesis contains a description of the optimization system implementation and discusses the experiments with the system.

Το κύριο στοιχείο της διατριβής είναι οι Εξελικτικοί Αλγόριθμοι. Στο πρώτο μέρος παρουσιάζονται οι Μιμιδικοί Αλγόριθμοι. Οι Μιμιδικοί Αλγόριθμοι είναι υβριδικά σχήματα που συνδυάζουν τους Εξελιτκικούς Αλγορίθμους με μεθόδους τοπικής αναζήτησης. Οι Μιμιδικοί Αλγόριθμοι συγκρίθηκαν με τους Εξελικτικούς Αλγορίθμους σε πληθώρα προβλημάτων ολικής βελτιστοποίησης και είχαν καλύτερα αποτελέσματα. Στο δεύτερο μέρος μελετήθηκαν προβλήματα μη γραμμικής δυναμικής. Αυτά ήταν η εκτίμηση της περιοχής ευστάθειας διατηρητικών απεικονίσεων, η ανίχνευση συντονισμών και ο υπολογισμός περιοδικών τροχιών. Τα αποτελέσματα ήταν ικανοποιητικά. / The main objective of the thesis was the study of Evolutionary Algorithms. At the first part, Memetic Algorithms were introduced. Memetic Algorithms are hybrid schemes that combine Evolutionary Algorithms and local search methods. Memetic Algorithms were compared to Evolutionary Algorithms in various problems of global optimization and they had better performance. At the second part, problems from nonlinear dynamics were studied. These were the estimation of the stability region of conservative maps, the detection of resonances and the computation of periodic orbits. The results were satisfactory.

Local search algorithms have been proved to be effective in refining solutions that have been found by other algorithms. Evolutionary algorithms, in particular global search algorithms, have shown to be successful in producing approximate solutions for optimisation and classification problems in acceptable computation times. A relatively new method, memetic algorithms, uses local search to refine the approximate solutions produced by global search algorithms. This thesis develops such a memetic algorithm. The global search algorithm used as part of the new memetic algorithm is a genetic program that implements the building block hypothesis by building simplistic decision trees representing valid solutions, and gradually increases the complexity of the trees. The specific building block hypothesis implementation is known as the building block approach to genetic programming, BGP. The effectiveness and efficiency of the new memetic algorithm, which combines the BGP algorithm with a local search algorithm, is demonstrated. / Dissertation (MSc)--University of Pretoria, 2006. / Computer Science / unrestricted

The automatic recognition of individuals through the iris characteristics is an e±cient biometric technique that is widely studied and applied around the world. Many image processing stages are necessary to make possible the representation and the interpretation of the iris information. This work presents the state of the art in iris recognition systems where the most re- markable works and the di®erent techniques applied to perform each process- ing stage are quoted. The implementations of each processing stage using traditional techniques are presented and, afterwards, two innovator methods are proposed with the common objective of bringing bene¯t to the system. The ¯rst processing stage should be the localization of the iris region in an eye image. The ¯rst method proposed in this work presents an algorithm to achieve the iris localization through the utilization of the called Memetic Algorithms. The new method is compared to a classical method and the obtained results show advantages concerning e±ciency and processing time. In another processing stage there must be a pixels sampling from the iris region, from where the information used to di®erentiate the individuals is extracted. Traditionally, this sampling process is accomplished in an uni- form way along the whole iris region. It is proposed a pre-processing method which suggests a non uniform pixels sampling from the iris region with the objective of selecting the group of pixels which carry more information about the iris structure. The search for this group of pixels is done through Ge- netic Algorithms. The application of the new method improves the e±ciency of the system and also, allows the generation of smaller templates. In this work, a study on the called Active Shape Models is also accomplished and its application to perform the iris region segmentation is evaluated. To execute the simulations and the evaluation of the methods, it was used two public and free iris images database: UBIRIS database and MMU database. / O reconhecimento automático de pessoas utilizando-se características da íris é uma eficiente técnica biométrica que está sendo largamente estudada e aplicada em todo o mundo. Diversas etapas de processamento são necessárias para tornar possível a representação e a interpretação da informação contida na íris. Neste trabalho é apresentado o estado da arte de sistemas de reconhecimento de íris onde são citados os trabalhos de maior destaque e as diferentes técnicas empregadas em cada etapa de processamento. São apresentadas implementações de cada etapa de processamento utilizando técnicas tradicionais e, posteriormente, são propostos dois métodos inovadores que têm o objetivo comum de trazer benefícios ao sistema. A primeira etapa de processamento é a localização da região da íris na imagem. O primeiro método proposto neste trabalho apresenta um algoritmo para realizar a localização da íris utilizando os chamados Algoritmos Meméticos. O novo método é comparado a um método clássico e os resultadosnobtidos demonstram vantagens no que diz respeito à eficiência e ao tempo de processamento. Em uma outra etapa de processamento deve haver uma amostragem de pixels na região da íris, de onde são retiradas as informações utilizadas para diferenciar os indivíduos. Tradicionalmente, esta amostragem é realizada de maneira uniforme ao longo de toda a região da íris. É proposto um método de pré-processamento que sugere uma amostragem não uniforme de pixels na região da íris com o objetivo de selecionar o conjunto de pixels que carregam mais informações da estrutura da íris. A busca por esse conjunto de pixels é realizada utilizando-se Algoritmos Genéticos. A aplicação deste novo método aumenta a eficiência do sistema e ainda possibilita a geração de templates binários menores. Neste trabalho é realizado, ainda, um estudos dos chamados Active Shape Models e a sua aplicação para segmentar a região da íris é avaliada. Para a simulação e avaliação dos métodos, foram utilizados dois bancos de imagens de íris públicos e gratuitos: o banco de imagens UBIRIS e o banco de imagens MMU. / Doutor em Ciências

Made available in DSpace on 2014-12-17T14:53:23Z (GMT). No. of bitstreams: 1 FabioFCF.pdf: 875120 bytes, checksum: 089fb9e8e722351411a9dbd3d86bbef4 (MD5) Previous issue date: 2006-05-19 / The Combinatorial Optimization is a basic area to companies who look for competitive advantages in the diverse productive sectors and the Assimetric Travelling Salesman Problem, which one classifies as one of the most important problems of this area, for being a problem of the NP-hard class and for possessing diverse practical applications, has increased interest of researchers in the development of metaheuristics each more efficient to assist in its resolution, as it is the case of Memetic Algorithms, which is a evolutionary algorithms that it is used of the genetic operation in combination with a local search procedure. This work explores the technique of Viral Infection in one Memetic Algorithms where the infection substitutes the mutation operator for obtaining a fast evolution or extinguishing of species (KANOH et al, 1996) providing a form of acceleration and improvement of the solution . For this it developed four variants of Viral Infection applied in the Memetic Algorithms for resolution of the Assimetric Travelling Salesman Problem where the agent and the virus pass for a symbiosis process which favored the attainment of a hybrid evolutionary algorithms and computational viable / A Otimiza??o Combinat?ria ? uma ?rea fundamental para empresas que buscam vantagens competitivas nos diversos setores produtivos, e o Problema do Caixeiro Viajante Assim?trico, o qual se classifica como um dos mais importantes problemas desta ?rea, devido a ser um problema da classe NP-dif?cil e tamb?m por possuir diversas aplica??es pr?ticas, tem despertado interesse de pesquisadores no desenvolvimento de Metaheur?sticas cada vez mais eficientes para auxiliar na sua resolu??o, como ? o caso do Algoritmo Mem?tico, o qual ? um algoritmo evolutivo que se utiliza dos operadores gen?ticos em combina??o com um procedimento de busca local. Este trabalho explora a t?cnica de Infec??o Viral em um Algoritmo Mem?tico, onde a infec??o substitui o operador de muta??o por conseguir uma r?pida evolu??o ou extin??o de esp?cies (KANOH et al., 1996), proporcionando uma forma de acelera??o e melhoria da solu??o. Para isto se desenvolveu quatro variantes de Infec??o Viral aplicadas no Algoritmo Mem?tico para resolu??o do Problema do Caixeiro Viajante Assim?trico, onde o agente e o v?rus passam por um processo de Simbiose, as quais favoreceram a obten??o de um algoritmo evolutivo h?brido e computacionalmente vi?vel