Facing global and dynamic competition environment, companies have to collaborate with other companies instead of struggle alone to optimize performance of supply chain. In a distributed supply chain structure, it is an important issue for companies to coordinate seamlessly to effectively fulfill customer orders. In this thesis, we seek to propose a fast and flexible method to solve the order fulfillment scheduling conflicts among partners in a supply chain. Due to the risk of exposing trade secrets and the cost of gathering information, the centralized constraint satisfaction mechanism is infeasible to handle distributed scheduling problem in real world environment. Moreover, the distributed constraints satisfaction model just focuses on finding a globally executable order fulfillment schedule. Therefore, we propose an agent-based distributed coordination mechanism that integrates negotiation with generic algorithm. We chose the mold manufacturing industry as an example and conducted experiments to evaluate the performance of the proposed mechanism and to compare with other benchmark methods proposed by researchers prior to this study. The experimental results indicate that the distributed coordination mechanism we proposed is a feasible approach to solve the order fulfillment scheduling conflicts in outsourcing activities in a supply chain.

The adaptiveness provides uniquely identifying and monitoring the learner&rsquo / s learning activities according to his/her respective profile. The adaptive intelligent learning management systems (AILMS) help a wide range of students to achieve their learning goals effectively by delivering knowledge in an adaptive or individualized style through online learning settings. This study presents a multi-agent system, called MODA, developed to provide adaptiveness in learning management systems (LMS). A conceptual framework for adaptive learning systems is proposed for this purpose. The framework is based on the idea that adaptiveness is the best matching between the learner profile and the course content profile. The learning styles of learners and the content type of learning material are used to match the learner to the most suitable content. The thesis covers the pedagogical framework applied in MODA, the technical and multi-agent architectures of MODA, the TCP-IP based protocol providing communication between MODA and LMS, and a sample application of the system to an open source learning management system, OLAT. The study also discusses the possibilities of future interests.

In the domain of the complex control problems for agents, neuroevolution, i.e. artificial evolution of neural networks, methods have been continuously shown to offer high performance solutions which may be unpredictable by external controller design. Recent studies have proved that these methods can also be successfully applied for cooperative multi-agent systems to evolve the desired team behavior. For a given task which may benefit from both cooperation and behavioral specialization, the genetic diversity of the team members may have important effects on the team performance. In this thesis, the single prey hunting problem is chosen as the case, where the performance of the evolved predator teams with distinct genetic similarity levels are systematically examined. For this purpose, three similarity levels, namely homogeneous, partially heterogeneous and heterogeneous, are adopted and analyzed in various problem-specific and algorithmic settings. Our similarity levels differ from each other in terms of the number of groups of identical agents in a single predator team, where identicalness of two agents refers to the fact that both have the same synaptic weight vector in their neural network controllers. On the other hand, the problem-specific conditions comprise three different fields of vision for predators, whereas algorithmic settings refer to varying number of individuals in the populations, as well as two different selection levels such as team and group levels. According to the experimental results within a simulated grid environment, we show that different genetic similarity level-field of vision-algorithmic setting combinations beget different performance results.

Along with the growth of electronic commerce has come an interest in developing autonomous trading agents. Often, such agents must interact directly with other market participants, and so the behavior of these participants must be taken into account when designing agent strategies. One common approach is to build a model of the market, but this approach requires the use of historical market data, which may not always be available. This dissertation addresses such a case: that of an agent entering a new market in which it has no previous experience. While the agent could adapt by learning about the behavior of other market participants, it would need to do so in an online fashion. The agent would not necessarily have to learn from scratch, however. If the agent had previous experience in similar markets, it could use this experience to tailor its learning approach to its particular situation. This dissertation explores methods that a trading agent could use to take advantage of previous market experience when adapting to a new market. Two distinct learning settings are considered. In the first, an agent acting as an auctioneer must adapt the parameters of an auction mechanism in response to bidder behavior, and a reinforcement learning approach is used. The second setting concerns agents that must adapt to the behavior of competitors in two scenarios from the Trading Agent Competition: supply chain management and ad auctions. Here, the agents use supervised learning to model the market. In both settings, methods of adaptation can be divided into four general categories: i) identifying the most similar previously encountered market, ii) learning from the current market only, iii) learning from the current market but using previous experience to tune the learning algorithm, and iv) learning from both the current and previous markets. The first contribution of this dissertation is the introduction and experimental validation of a number of novel algorithms for market adaptation fitting these categories. The second contribution is an exploration of the degree to which the quantity and nature of market experience impact the relative performance of methods from these categories. / text

<p>This dissertation presents a novel distributed optimal control (DOC) problem formulation that is applicable to multiscale dynamical systems comprised of numerous interacting systems, or agents, that together give rise to coherent macroscopic behaviors, or coarse dynamics, that can be modeled by partial differential equations (PDEs) on larger spatial and time scales. The DOC methodology seeks to obtain optimal agent state and control trajectories by representing the system's performance as an integral cost function of the macroscopic state, which is optimized subject to the agents' dynamics. The macroscopic state is identified as a time-varying probability density function to which the states of the individual agents can be mapped via a restriction operator. Optimality conditions for the DOC problem are derived analytically, and the optimal trajectories of the macroscopic state and control are computed using direct and indirect optimization algorithms. Feedback microscopic control laws are then derived from the optimal macroscopic description using a potential function approach.</p><p>The DOC approach is demonstrated numerically through benchmark multi-agent trajectory optimization problems, where large systems of agents were given the objectives of traveling to goal state distributions, avoiding obstacles, maintaining formations, and minimizing energy consumption through control. Comparisons are provided between the direct and indirect optimization techniques, as well as existing methods from the literature, and a computational complexity analysis is presented. The methodology is also applied to a track coverage optimization problem for the control of distributed networks of mobile omnidirectional sensors, where the sensors move to maximize the probability of track detection of a known distribution of mobile targets traversing a region of interest (ROI). Through extensive simulations, DOC is shown to outperform several existing sensor deployment and control strategies. Furthermore, the computation required by the DOC algorithm is proven to be far reduced compared to that of classical, direct optimal control algorithms.</p> / Dissertation

The complexity of the strategic decision making environments, in which busi- nesses and governments live in, makes such decisions more and more difficult to make. People and organizations with access to the best known decision support modelling and analysis tools and methods cannot seem to benefit from such re- sources. We argue that the reason behind the failure of most current decision and game theoretic methods is that these methods are made to deal with operational and tactical decisions, not strategic decisions. While operational and tactical decisions are clear and concise with limited scope and short-term implications, allowing them to be easily formalized and reasoned about, strategic decisions tend to be more gen- eral, ill-structured, complex, with broader scope and long-term implications. This research work starts with a review of the current dominant modelling and analysis approaches, their strengths and shortcomings, and a look at how pioneers in the field criticize these approaches as restrictive and unpractical. Then, the work goes on to propose a new paradigm shift in how strategic decisions and conflicts should be modelled and analyzed. Constrained Rationality is a formal qualitative framework, with a robust method- ological approach, to model and analyze ill-structured strategic single and multi- agent decision making situations and conflicts. The framework brings back the strategic decision making problem to its roots, from being an optimization/efficiency problem about evaluating predetermined alternatives to satisfy predetermined pref- erences or utility functions, as most current decision and game theoretic approaches treats it, to being an effectiveness problem of: 1) identifying and modelling explic- itly the strategic and conflicting goals of the involved agents (also called players and decision makers in our work), and the decision making context (the external and internal constraints including the agents priorities, emotions and attitudes); 2) finding, uncovering and/or creating the right set of alternatives to consider; and then 3) reasoning about the ability of each of these alternatives to satisfy the stated strategic goals the agents have, given their constraints. Instead of assuming that the agents’ alternatives and preferences are well-known, as most current decision and game theoretic approaches do, the Constrained Rationality framework start by capturing and modelling clearly the context of the strategic decision making situation, and then use this contextual knowledge to guide the process of finding the agents’ alternatives, analyzing them, and choosing the most effective one. The Constrained Rationality framework, at its heart, provides a novel set of modelling facilities to capture the contextual knowledge of the decision making sit- uations. These modelling facilities are based on the Viewpoint-based Value-Driven - Enterprise Knowledge Management (ViVD-EKM) conceptual modelling frame- work proposed by Al-Shawa (2006b), and include facilities: to capture and model the goals and constraints of the different involved agents, in the decision making situation, in complex graphs within viewpoint models; and to model the complex cause-effect interrelationships among theses goals and constraints. The framework provides a set of robust, extensible and formal Goal-to-Goal and Constraint-to Goal relationships, through which qualitative linguistic value labels about the goals’ op- erationalization, achievement and prevention propagate these relationships until they are finalized to reflect the state of the goals’ achievement at any single point of time during the situation. The framework provides also sufficient, but extensible, representation facilities to model the agents’ priorities, emotional valences and attitudes as value properties with qualitative linguistic value labels. All of these goals and constraints, and the value labels of their respective value properties (operationalization, achievement, prevention, importance, emotional valence, etc.) are used to evaluate the different alternatives (options, plans, products, product/design features, etc.) agents have, and generate cardinal and ordinal preferences for the agents over their respective alternatives. For analysts, and decision makers alike, these preferences can easily be verified, validates and traced back to how much each of these alternatives con- tribute to each agent’s strategic goals, given his constraints, priorities, emotions and attitudes. The Constrained Rationality framework offers a detailed process to model and analyze decision making situations, with special paths and steps to satisfy the spe- cific needs of: 1) single-agent decision making situations, or multi-agent situations in which agents act in an individualistic manner with no regard to others’ current or future options and decisions; 2) collaborative multi-agent decision making situ- ations, where agents disclose their goals and constraints, and choose from a set of shared alternatives one that best satisfy the collective goals of the group; and 3) adversarial competitive multi-agent decision making situations (called Games, in gamete theory literature, or Conflicts, in the broader management science litera- ture). The framework’s modelling and analysis process covers also three types of con- flicts/games: a) non-cooperative games, where agents can take unilateral moves among the game’s states; b) cooperative games, with no coalitions allowed, where agents still act individually (not as groups/coalitions) taking both unilateral moves and cooperative single-step moves when it benefit them; and c) cooperative games, with coalitions allowed, where the games include, in addition to individual agents, agents who are grouped in formal alliances/coalitions, giving themselves the ability to take multi-step group moves to advance their collective position in the game. ....

À venir / The customary regime of NRM in Bhutan faces greater challenges from economic development, rapid transformation of social values, local institutions and traditional perceptions on NR. Although Bhutan is projected rich in water resource for hydropower potentials, water for agriculture and domestic use is fast becoming scarcer and highly contested. As the water becomes scarce the people living in highlands are most severely affected. A detail diagnostic study of two communities, Lingmuteychu depedent of irrigated rice and Kengkhar depedent on dryland farming presents two situations of water resource issues. In Lingmuteychu the conflict of irrigation water sharing for cultivation of rice among seven villages has been inflicting restentment in all aspect of society. In contrary, in Kengkhar has been facing drinking water scarcity as the natural spring ponds dry, which forces people to walk from more than five hours to fetch water from the river. In these two sites Companion Modelling was applied to enhance understanding of water resource management dynamics and improve shared communication and learning to facilitate adaptative management dynamics and improve shared communication and learning to facilitate adaptative management strategies. The study provides as comparison of the process followerd in two sites with analysis of impacts and effects from the process. The research illustrates fow ComMod process help develop trust and and commitment in the conflicting community and pave pathways to develop social capital for adaptive management of water resource. The process was able to foster shared learning and co-construct collective actions which were implementable. The research also revealed the important role of the researchers in furthering and sustaining newly achieved cooperation. The ABMs provided an opportunity to re-create different water resource management strategies which could be used as options for the community. The research also demonstrates the feasibility of applying the ComMod approach in different fields of NRM.

La confiance et l'incertitude sont deux aspects importants des systèmes distribués. Par exemple, de multiples sources d'information peuvent fournir le même type d'information. Cela pose le problème de sélectionner la source la plus fiable et de résoudre l'incohérence dans l'information disponible. Gérer de front la confiance et l'incertitude constitue un problème complexe et nous développons à travers cette thèse, une solution pour y répondre. La confiance et l'incertitude sont intrinsèquement liés. La confiance concerne principalement les sources d'information alors que l'incertitude est une caractéristique de l'information elle-même. En l'absence de mesures de confiance et d'incertitude, un système doit généralement faire face à des problèmes tels que l'incohérence et l'incertitude. Pour aborder ce point, nous émettons l'hypothèse que les sources dont les niveaux de confiance sont élevés produiront de l'information plus fiable que les sources dont les niveaux de confiance sont inférieurs. Nous utilisons ensuite les mesures de confiance des sources pour quantifier l'incertitude dans l'information et ainsi obtenir des conclusions de plus haut niveau avec plus de certitude.Une tendance générale dans les systèmes distribués modernes consiste à intégrer des capacités de raisonnement dans les composants pour les rendre intelligents et autonomes. Nous modélisons ces composants comme des agents d'un système multi-agents. Les principales sources d'information de ces agents sont les autres agents, et ces derniers peuvent posséder des niveaux de confiance différents. De plus, l'information entrante et les croyances qui en découlent sont associées à un degré d'incertitude. Par conséquent, les agents sont confrontés à un double problème: celui de la gestion de la confiance sur les sources et celui de la présence de l'incertitude dans l'information. Nous illustrons cela avec trois domaines d'application: (i) la communauté intelligente, (ii) la collecte des déchets dans une ville intelligente, et (iii) les facilitateurs pour les systèmes de l'internet du futur (FIWARE - le projet européen n° 285248, qui a motivé la recherche sur nos travaux). La solution que nous proposons consiste à modéliser les composants de ces domaines comme des agents intelligents qui incluent un module de gestion de la confiance, un moteur d'inférence et un système de révision des croyances. Nous montrons que cet ensemble d'éléments peut aider les agents à gérer la confiance aux autres sources, à quantifier l'incertitude dans l'information et à l'utiliser pour aboutir à certaines conclusions de plus haut niveau. Nous évaluons finalement notre approche en utilisant des données à la fois simulées et réelles relatives aux différents domaines d'application. / Trust and uncertainty are two important aspects of many distributed systems. For example, multiple sources of information can be available for the same type of information. This poses the problem to select the best source that can produce the most certain information and to resolve incoherence amongst the available information. Managing trust and uncertainty together forms a complex problem and through this thesis we develop a solution to this. Trust and uncertainty have an intrinsic relationship. Trust is primarily related to sources of information while uncertainty is a characteristic of the information itself. In the absence of trust and uncertainty measures, a system generally suffers from problems like incoherence and uncertainty. To improve on this, we hypothesize that the sources with higher trust levels will produce more certain information than those with lower trust values. We then use the trust measures of the information sources to quantify uncertainty in the information and thereby infer high level conclusions with greater certainty.A general trend in the modern distributed systems is to embed reasoning capabilities in the end devices to make them smart and autonomous. We model these end devices as agents of a Multi Agent System. Major sources of beliefs for such agents are external information sources that can possess varying trust levels. Moreover, the incoming information and beliefs are associated with a degree of uncertainty. Hence, the agents face two-fold problems of managing trust on sources and presence of uncertainty in the information. We illustrate this with three application domains: (i) The intelligent community, (ii) Smart city garbage collection, and (iii) FIWARE : a European project about the Future Internet that motivated the research on this topic. Our solution to the problem involves modelling the devices (or entities) of these domains as intelligent agents that comprise a trust management module, an inference engine and a belief revision system. We show that this set of components can help agents to manage trust on the other sources and quantify uncertainty in the information and then use this to infer more certain high level conclusions. We finally assess our approach using simulated and real data pertaining to the different application domains.

[pt] Nos últimos anos a área de Sistemas Multi-Agentes (SMAs) vem apresentando um crescimento acelerado. Novas técnicas e ferramentas estão surgindo e a cada dia aumenta o número de pessoas dedicadas ao tema. Muitas metodologias para o desenvolvimento de sistemas multi-agentes têm sido propostas. No entanto, a maioria delas dedica-se principalmente à fase de análise dos sistemas. Este trabalho propõe um método para a implementação de SMAs utilizando componentes de software. Na fase de análise e projeto foi utilizada a linguagem ANote, que compreende sete diagramas com o objetivo de modelar os aspectos de um SMA com uma notação própria para agentes e com diversas visões do sistema. Em seguida foi proposto um modelo de implementação dos agentes baseado em componentes e são descritos os mapeamentos necessários para transformar a modelagem do SMA em um sistema implementado. Para validar o modelo é apresentado um Estudo de Caso como prova de conceito das idéias presentes nesta proposta. O Estudo de Caso consiste em um mercado virtual onde os agentes são responsáveis pela compra e venda de produtos. Esta implementação utiliza o CORBA Component Model (CCM) e uma linguagem para comunicação entre agentes, a FIPA-ACL. / [en] In the past few years, the Multi-Agents Systems (MAS) area has presented an accelerated growth. New techniques and tools are constantly being proposed and the number of specialists dedicated to this subject is increasing. Many methodologies have been published to support the development of multi-agent systems. However, most of them concentrate only on the system analysis phase. This work proposes a method to implement MASs using software components. During the analysis and design phases, the ANote language was used. It contains seven diagrams that model different aspects of a MAS and a proper notation for describing agents and different views of the system. An agent implementation model based on components is proposed and the mappings from the MAS elements to the system implementation are described. To validate the model, a Case Study is presented using the concepts described in this proposal. The Case Study consists of a virtual marketplace where agents are responsible for buying and selling products. The implementation uses the CORBA Component Model (CCM) and a language for agent communication called FIPA- ACL.

[pt] A mineração de dados é um tema em alta que atrai pesquisadores de diferentes áreas, como bancos de dados, aprendizado de máquina e sistemas multiagentes. Como consequência do crescimento do volume de dados, há uma necessidade crescente de obter conhecimento desses grandes conjuntos de dados que são muito difíceis de manipular e processar com os métodos tradicionais. Os agentes de software podem desempenhar um papel significativo ao executar processos de mineração de dados de maneira mais eficiente. Por exemplo, eles podem trabalhar para realizar seleção, extração, pré-processamento e integração de dados, bem como mineração paralela, distribuída ou de múltiplas fontes. Este trabalho propõe uma abordagem (na forma de um framework) que usa agentes de software para gerenciar processos de mineração de dados. Para testar sua aplicabilidade, utilizamos vários conjuntos de dados relacionados ao domínio de saúde, representando alguns cenários de uso (hipotireoidismo, diabetes e arritmia). / [en] Data mining is a hot topic that attracts researchers from different areas, such as databases, machine learning, and multi-agent systems. As a consequence of the growth of data volume, there is a growing need to obtain knowledge from these large data sets that are very difficult to handle and process with traditional methods. Software agents can play a significant role performing data mining processes in ways that are more efficient. For instance, they can work to perform selection, extraction, preprocessing and integration of data as well as parallel, distributed, or multisource mining. This work proposes an approach (in the form of a framework) that uses software agents to manage data mining processes. In order to test its applicability, we use several data sets related to health care domain representing some usage scenarios (hypothyroidism, diabetes and arrhythmia).