This dissertation is an attempt of expanding the domain of game theory into the sphere of evolving, potentially non-equilibrium systems. We especially focus our attention on studying the effects of local interactions, using automata networks as a modelling tool. The Chapters 2 and 3 of this dissertation concentrate on applications of the local nature of interactions and rely on automata networks as an investigating and modelling tool for game theory. Chapter 2 is devoted to cooperation and to a smaller extent to the endogenous formation of links between the agents. Chapter 3 is investigating the deterministic and stochastic best response play when interactions are local. / Ph. D.

As recentes pesquisas na área de Gestão de Processos de Negócio (GPN) vêm contribuindo para aumentar a eficiência nas organizações. A GPN pode ser compreendida como o conjunto de métodos, técnicas e ferramentas computacionais desenvolvidas para amparar os processos de negócios. Tipicamente, a GPN é fundamentada por modelos de processos. Esses modelos, além de permitirem a automação da configuração e execução, aumentam a capacidade de análise dos processos de negócio. Apesar de auxiliar os especialistas de negócio nas diferentes fases envolvidas no ciclo de vida de um processo de negócio (projeto, configuração, implantação/execução e a análise), os modelos definidos em linguagens específicas de domínio, como a BPMN (Business Process Model and Notation), não são os mais apropriados para amparar a fase de análise. De formal geral, esses modelos não possuem uma semântica operacional formalmente definida (o que limita o seu uso para a verificação e validação dos processos) e nem mecanismos para quantificar o comportamento modelado (o que impossibilita a análise de desempenho). Neste trabalho de doutorado, nós desenvolvemos um arcabouço que ampara e automatiza os principais passos envolvidos na análise de desempenho de processos de negócio via modelagem analítica. Nós estudamos a viabilidade da aplicação de três formalismos Markovianos na modelagem de processos de negócio: as Redes de Petri Estocásticas, as Álgebras de Processo Estocásticas e as Redes de Autômatos Estocásticos (SAN, do inglês Stochastic Automata Networks). Escolhemos SAN como formalismo base para o método proposto neste trabalho. Nosso arcabouço é constituído por: (i) uma notação para enriquecer modelos de processos de negócio descritos em BPMN com informações sobre o seu gerenciamento de recursos, e (ii) um algoritmo que faz a conversão automática desses modelos não-formais de processos para modelos estocásticos em SAN. Com isso, somos capazes de capturar o impacto causado pela contenção de recursos no desempenho de um processo de negócio. A partir de um modelo em SAN gerado com o nosso arcabouço, podemos predizer variados índices de desempenho que são boas aproximações para o desempenho esperado do processo de negócio no mundo real. / Recent results in the research field of Business Process Management (BPM) are contributing to improve efficiency in organizations. BPM can be seen as a set of methods, techniques and tools developed to support business processes in their different requirements. Usually, the BPM techniques are based on a process model. In addition to enabling automated process configuration and execution, these models also increase the analizability of business processes. Despite being able to support business specialists in different phases of the life cycle of a business process (design, configuration, execution, and analysis), the models created in domain-specific languages, such as BPMN (Business Process Model and Notation), are not the most appropriated ones to support the analysis phase. Generally, these models have neither a formally defined operational semantics (which hinders their use for verification and validation), nor mechanisms to quantify the modeled behavior (which hinders their use for performance analysis). In this PhD research, we developed a framework to support and to automatize the main steps involved in the analytical modeling of business processes aiming performance evaluation. We studied the viability of applying three Markovian formalisms in business process modeling: Stochastic Petri Nets, Stochastic Process Algebras and Stochastic Automata Networks (SAN). We have chosen SAN to support the method proposed in this work. Our framework is composed of: (i) a notation to enrich BPMN business process models with information concerning the associated resource management and (ii) an algorithm that automatically converts these non-formal business process models in SAN stochastic models. With this, we are able to capture the impact caused by resource contention in the performance of a business process. From a model generated through our framework, we are able to extract varied performance indices that are good approximations for the expected process performance in the real world.

As recentes pesquisas na área de Gestão de Processos de Negócio (GPN) vêm contribuindo para aumentar a eficiência nas organizações. A GPN pode ser compreendida como o conjunto de métodos, técnicas e ferramentas computacionais desenvolvidas para amparar os processos de negócios. Tipicamente, a GPN é fundamentada por modelos de processos. Esses modelos, além de permitirem a automação da configuração e execução, aumentam a capacidade de análise dos processos de negócio. Apesar de auxiliar os especialistas de negócio nas diferentes fases envolvidas no ciclo de vida de um processo de negócio (projeto, configuração, implantação/execução e a análise), os modelos definidos em linguagens específicas de domínio, como a BPMN (Business Process Model and Notation), não são os mais apropriados para amparar a fase de análise. De formal geral, esses modelos não possuem uma semântica operacional formalmente definida (o que limita o seu uso para a verificação e validação dos processos) e nem mecanismos para quantificar o comportamento modelado (o que impossibilita a análise de desempenho). Neste trabalho de doutorado, nós desenvolvemos um arcabouço que ampara e automatiza os principais passos envolvidos na análise de desempenho de processos de negócio via modelagem analítica. Nós estudamos a viabilidade da aplicação de três formalismos Markovianos na modelagem de processos de negócio: as Redes de Petri Estocásticas, as Álgebras de Processo Estocásticas e as Redes de Autômatos Estocásticos (SAN, do inglês Stochastic Automata Networks). Escolhemos SAN como formalismo base para o método proposto neste trabalho. Nosso arcabouço é constituído por: (i) uma notação para enriquecer modelos de processos de negócio descritos em BPMN com informações sobre o seu gerenciamento de recursos, e (ii) um algoritmo que faz a conversão automática desses modelos não-formais de processos para modelos estocásticos em SAN. Com isso, somos capazes de capturar o impacto causado pela contenção de recursos no desempenho de um processo de negócio. A partir de um modelo em SAN gerado com o nosso arcabouço, podemos predizer variados índices de desempenho que são boas aproximações para o desempenho esperado do processo de negócio no mundo real. / Recent results in the research field of Business Process Management (BPM) are contributing to improve efficiency in organizations. BPM can be seen as a set of methods, techniques and tools developed to support business processes in their different requirements. Usually, the BPM techniques are based on a process model. In addition to enabling automated process configuration and execution, these models also increase the analizability of business processes. Despite being able to support business specialists in different phases of the life cycle of a business process (design, configuration, execution, and analysis), the models created in domain-specific languages, such as BPMN (Business Process Model and Notation), are not the most appropriated ones to support the analysis phase. Generally, these models have neither a formally defined operational semantics (which hinders their use for verification and validation), nor mechanisms to quantify the modeled behavior (which hinders their use for performance analysis). In this PhD research, we developed a framework to support and to automatize the main steps involved in the analytical modeling of business processes aiming performance evaluation. We studied the viability of applying three Markovian formalisms in business process modeling: Stochastic Petri Nets, Stochastic Process Algebras and Stochastic Automata Networks (SAN). We have chosen SAN to support the method proposed in this work. Our framework is composed of: (i) a notation to enrich BPMN business process models with information concerning the associated resource management and (ii) an algorithm that automatically converts these non-formal business process models in SAN stochastic models. With this, we are able to capture the impact caused by resource contention in the performance of a business process. From a model generated through our framework, we are able to extract varied performance indices that are good approximations for the expected process performance in the real world.

Submitted by Maicon Juliano Schmidt (maicons) on 2015-03-23T14:28:22Z No. of bitstreams: 1 Luan Carlos Nesi.pdf: 100100607 bytes, checksum: 6012e0f177d7b8f3807de72ff7d98315 (MD5) / Made available in DSpace on 2015-03-23T14:28:22Z (GMT). No. of bitstreams: 1 Luan Carlos Nesi.pdf: 100100607 bytes, checksum: 6012e0f177d7b8f3807de72ff7d98315 (MD5) Previous issue date: 2014-03-27 / Milton Valente / Nesse trabalho foi desenvolvido um modelo computacional de Hipermídia Adaptativa para geração de layouts de interface de aplicações. A pesquisa partiu de uma revisão sobre Hipermídia Adaptativa, com um apanhado sobre os conceitos e características dos métodos e técnicas de adaptação a fim de embasar seu desenvolvimento. Após, avaliou-se o uso das metaheurísticas Algoritmo Genético, Busca Tabu e Algoritmo Memético como as ferramentas de apoio no desenvolvimento do modelo. Na sequência, as Redes de Autômatos Estocásticos nortearam a modelagem do formalismo utilizado para a retenção de conhecimento. Dessas bases, foi desenvolvida a prova de conceito. Conseguinte, apresentam-se os experimentos realizados para validação. Os resultados obtidos pelo modelo foram de boa qualidade, indo ao encontro dos objetivos da pesquisa. Como decorrência deste trabalho, obteve-se um sistema capaz de gerar layouts, contemplando as características dos usuários e seus dispositivos, sendo capaz de acompanhar uma tendência de consumo de conteúdos não só mercadológica, mas também, social. / In this paper was developed a computational model of Adaptive Hypermedia for generation of interface layouts of applications. The research began with a review of Adaptive Hypermedia, with an overview of the concepts and characteristics of the methods and adaptation techniques in order to base its development. After, we evaluated the use of metaheuristic Genetic Algorithm, Tabu Search, and Memetic Algorithm as support tools in the development of the model. Following, the Stochastic Automata Networks guided the modeling of the formalism used for knowledge retention. These bases, the proof of concept were developed. Therefore, we present the experiments to validate. The obtained results by the model were of good quality, meeting the research objectives. As results of this work, we obtained a system capable to generate layouts, considering the characteristics of the users and their devices, being able to follow a trend of content consumption not only marketing, but also social.

We extend the study of learning and generalization in feed forward Boolean networks to random Boolean networks (RBNs). We explore the relationship between the learning capability and the network topology, the system size, the training sample size, and the complexity of the computational tasks. We show experimentally that there exists a critical connectivity Kc that improves the generalization and adaptation in networks. In addition, we show that in finite size networks, the critical K is a power-law function of the system size N and the fraction of inputs used during the training. We explain why adaptation improves at this critical connectivity by showing that the network ensemble manifests maximal topological diversity near Kc. Our work is partly motivated by self-assembled molecular and nanoscale electronics. Our findings allow to determine an automata network topology class for efficient and robust information processing.

Cette thèse s'intéresse aux évènements et aux ordonnancements d'évènements se produisant au sein de réseaux d'éléments conceptuels prédéterminés. Dans ces réseaux, les éléments, appelés plutôt "automates", s'incitent les uns les autres à changer d'état en accord avec des règles prédéfinies qui, précisément, définissent le (fonctionnement du) réseau. Lorsqu'un automate se conforme effectivement aux influences qu'il reçoit de la part des autres, on dit que son état est mis à jour. Les évènements élémentaires considérés sont les changements d'états des automates. Définir un mode de mise à jour pour l'ensemble des automates d'un réseau permet de sélectionner certains évènements parmi l'ensemble de ceux qui sont a priori possibles. Cela permet aussi d'organiser et d'ordonner les évènements les uns par rapport aux autres de façon, par exemple, à imposer que des évènements indépendants se produisent simultanément ou simplement, de manière assez rapprochée pour qu'aucun autre événement ne puisse se produire pendant leur occurrence. Informellement, les modes de mise à jour peuvent donc être interprétés comme l'expression d'influences extérieures au réseau interdisant certains changements, ou alors comme la formalisation d'une version relâchée et relative de l'écoulement de temps. Cette thèse propose d'étudier leur influence sur le comportement des réseaux. Et afin de distinguer cette influence de celle de la structure des réseaux, elle commence par mettre en évidence le rôle de certains motifs structurels. Après ça, elle s'intéresse en particulier à l'information "encodée" dans une séquence de mises à jour et à l'impact du synchronisme dans celles-ci. / This thesis is concerned with the events and the organisation of events that take place within networks of abstract predetermined elements called "automata". In these networks, automata incite one another to switch states in agreement with predefined rules which, precisely, define the net-work. When an automaton effectively conforms to the influences it receives from others, its state is said to be updated. The elementary events that are considered here are thus automata state changes. To define an update mode for all the automata of a network allows to select some events among all those that are a priori possible. It also allows to organise and order the events relatively so as to impose, for example, that independent events occur simultaneously or so that simply, they happen close enough to disallow the occurrence of any other events in between. Informally, update modes can be interpreted as the expressions of influences incoming from outside the network, forbidding certain changes, or else, as the formalisation of a relaxed and relative version of time flow. This thesis proposes to study their influences on network behaviours. And to distinguish their influences from that of network structures, it starts by highlighting the role of certain structural motives. After that, it explores in particular the information that is "encoded" in a sequence of updates as well as the general impact of synchronism in updates.