Orientador: José Geraldo Nery / Resumo: A busca por novos materiais, principalmente materiais que apresentem múltiplas aplicações, continua atraindo pesquisas ao redor do mundo. Dentre os materiais pesquisados com esse intuito, os compósitos ganham destaque. Compósitos são materiais mistos formados pela união de dois ou mais materiais diferentes com a finalidade de produzir um material novo com propriedades distintas, em relação aos seus materiais de origem. Os compósitos podem ser aplicados em diferentes áreas, entre elas encontram-se a remediação ambiental e a hemostasia. A remediação ambiental se faz necessária principalmente pelo fato de que muitas fontes hídricas acabam sendo poluídas por resíduos nocivos a saúde humana e animal, como metais pesados, e o consumo destas águas acaba causando sérias doenças. Devido a isso a busca por agentes capazes de retirar estes poluentes da água torna-se necessária. Outro grande problema de saúde publica são as hemorragias incontroladas que continuam sendo umas das principais causas de mortes no mundo. Neste contexto, o objetivo deste trabalho foi sintetizar um compósito que apresente a capacidade de atuar tanto como agente de remediação ambiental como agente hemostático. O compósito foi sintetizado utilizando zeólita e quitosana como suas matrizes. O material foi caracterizado por Difração de Raios-X; Espectroscopia de Infravermelho e Microscopia Eletrônica de Varredura. O potencial de ação ambiental foi testado analisando a absorção de cátions de cádmio e chumbo, presentes... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: The search for new materials, especially materials with multiple applications, continues to attract research from around the world. Among the materials researched for this purpose, composites stand out. Composites are mixed materials formed by joining two or more different materials in order to produce a new material with distinct properties relative to their source materials. Composites can be applied in different areas, including environmental remediation and hemostasis. Environmental remediation is necessary mainly due to the fact that many water sources end up being polluted by waste harmful to human and animal health, such as heavy metals, and the consumption of these waters ends up causing serious diseases. Because of this the search for agents capable of removing these pollutants from water becomes necessary. Another major public health problem is uncontrolled bleeding, which remains one of the leading causes of death worldwide. In this context, the objective of this work was to synthesize a composite that presents the ability to act as both environmental remediation agent and hemostatic agent. The composite was synthesized using zeolite and chitosan as their matrices. The material was characterized by X-ray diffraction; Infrared Spectroscopy and Scanning Electron Microscopy. The environmental action potential was tested by analyzing the absorption of cadmium and lead cations present in aqueous solution by the material, and the possible hemostatic application of the ma... (Complete abstract click electronic access below) / Mestre

The expansion and diffusion of mobile phones globally has resulted in the provision of financial transactional services over the existing mobile phone platforms, generally referred to as mobile money. The supply end of mobile money services is an important factor in the success of the financial transactions offering. This research assessed vulnerabilities in the mobile money supply network that are inherently related to the existence of the principal – agent problem and their implications on availability and access to the services. The research study was conducted using a qualitative approach. Qualitative information was collected through interviews guided by open – ended questionnaires. Thematic analysis approach was followed to systematically analyse the data and generate findings of the study. Agent transactional data was analysed to complement the findings from qualitative analysis The findings suggest that the principal agent problem permeates the mobile money delivery network mainly after businesses joining as agents and manifests as moral hazard. Moral hazard is the dominant feature of the principal – agent problem, with adverse selection very low. Drivers of moral hazard are demonstrated by the influences and demands of agents’ core businesses and challenges in agent monitoring and training. The existence of the principal – agent problem has limited or no implications on access and availability of services. However, overtime the combined vulnerabilities identified related to the principal agent problem are likely to manifest into risks that are likely to affect access and availability of mobile money services. Regulators, Mobile Network Operators and agent enterprises must collectively review monitoring approaches for mobile money service providers to address challenges identified and increase the effectiveness of monitoring. Service provision standards should be reviewed to suit the various business environments the services are provided within. Mobile Network Operators and agent enterprises need to institute stronger partnership arrangements that enhance ownership and obligations for all parties, in particular agent enterprises. Agreements must enable application of different mobile money delivery models suitable to meet the demands and requirements of the agents’ core businesses. Innovations such as Near Field Communication (NFC) can be integrated with Point of sale (POS) applications and mobile money platforms to reduce the administration burden on agents and human error. Such applications must consider the cost implications of adoption from the agents’ business perspective.

The financial markets have been started to get decentralized and even distributed. Consumers can now purchase stocks from their home computers without the use of a traditional broker. The dynamism and unpredictability of this domain which is continuously growing in complexity and also the giant volume of information which can affect this market, makes it one of the best potential domains to take advantage of agents. This thesis considers the main concerns of securities e-trading area in order to highlight advantages and disadvantages of multi-agent negotiating systems for online trading of securities comparing to single-agent systems. And then presents a multi-agent system design named MASTNA which considers both decision making and negotiating. The design seeks to improve the main concerns of securities e-trading such as speed, accuracy and handling complexities. MASTNA works over a distributed market and engages different types of agents in order to perform different tasks. For handling the negotiations MASTNA takes advantage of mobile negotiator agents with the purpose of handling parallel negotiations over an unreliable network (Internet).

Nous étudions dans cette thèse la formation et la dynamique des attitudes sociales à l'aide de la simulation multi agent. L'attitude peut se définir comme une évaluation globale d'un objet social sur la base d'informations cognitives ou affectives. Nos travaux s'inscrivent dans le cadre de la simulation sociale qui tente de reproduire informatiquement des phénomènes sociaux complexes à une échelle macroscopique, sur la base de la représentation des individus et de leurs interactions au niveau microscopique. Tandis que les approches existantes dans cette discipline font généralement abstraction des travaux en sciences humaines sur le sujet de l'attitude, nous proposons de suivre une approche psychomimétique en micro-fondant le modèle cognitif de nos agents sur des théories issues de psychologie sociale et des sciences cognitives. Ainsi, nous proposons un modèle de dynamique d'attitude combinant des théories issues de travaux en sciences humaines et sociales de la perception des individus, la communication inter-personnelle et médiatique, la révision de croyances, la réponse émotionnelle ainsi que le sentiment de surprise. Ce modèle a pour objectif de reproduire au niveau microscopique la dynamique des attitudes vis-à-vis d'acteurs qui effectuent des actions observées par la population. Nous avons procédé à une analyse fonctionnelle des différents composants du modèle sur des scénarii abstraits afin d'étudier les capacités de notre modèle, en particulier les phénomènes descriptibles tels que la diffusion de l'information, la résistance à la désinformation ou le processus de conformité. Le modèle a été appliqué dans le contexte des opérations militaires françaises de stabilisation en Afghanistan. L'objectif de cette expérience consiste à reproduire les sondages d'opinions vis-à-vis des Forces en présence, récoltés durant l'intervention, à partir d'un scénario militaire qui a été reconstitué en partenariat avec les officiers en charge des opérations de 2011 à 2012. Les résultats de simulations qui suivent un processus de calibration du modèle affichent une erreur inférieure à 3 points d'écart par rapport aux données réelles. Enfin, nous proposons une analyse microscopique des résultats en appliquant des techniques de classifications automatiques sur les individus afin d'expliquer les différentes tendances d'attitudes au sein de la population. / We study in this thesis the problem of social attitude formation and dynamics using multi agent simulation. The concept of attitude could be defined as a global evaluation of a social object, based on cognitive or affective information. Our works belongs to the field of social simulation which aims to reproduce in a virtual environment complex social phenomenon at a macroscopic level based on microscopic representations of individuals and their interactions. While existing approaches in this field rarely consider the results of studies in human sciences on the topic of attitude, we propose to follow a psychomimtic approach by micro-founding the cognitive model of our agents on human and social sciences' theories on individual's perception, inter-personal and media communication, belief revision, affective responses and the sentiment of unexpectedness. This model aims to reproduce at a microscopic level attitude dynamics toward actors who perpetuate actions witnessed by the individuals. We have proceeded to a functional analysis of the model's various components based on abstracts scenarios in order to study the capabilities of our model, and more precisely the describable phenomenon such as information diffusion, resistance to disinformation or the conformity process. The model has been applied in the context of French military operations of stabilisation in Afghanistan. The goal of this experience consists in reproducing opinion polls results of the locals toward the present Forces, collected during the intervention, based on a military scenario which has been recreated in partnership with officers who were in charge of operations between 2011 and 2012. Simulation results that follow a model calibration process show an error below 3 points of disparity compared to the real data. Finally, we propose a microscopic analysis of the results by applying automatic classification techniques on the simulated individuals in order to explain the multiple attitudes tendencies in the population.

Les interactions de l’équipe de travail et l’efficacité de celles-ci sont constitutives d’un système hiérarchique préétabli composé notamment de dyades verticales. Afin d’étudier cette dyade, ce travail doctoral s’intéresse aux rôles sociaux associés au statut de chacun des protagonistes de la dyade verticale (i.e., leader pour le chef et suiveur pour le subordonné) et considère le leadership.L’influence du suiveur et de ses comportements sur le leader et le leadership est de plus en plus considérée. Cependant, les comportements du suiveur influençant le leader restent inexplorés. Ce travail doctoral plurisdisciplinaire (i.e., psychologie et informatique affective) vise donc à mettre en évidence les comportements du suiveur influençant le leader ainsi que les mécamises sous-jacents à cette influence.Pour cela, nous nous sommes intéressés aux comportements non verbaux (CNV) du suiveur en tant qu’indices sociaux pouvant influencer le leader et ses comportements. Les principes de la cognition sociale sont appliqués dans cette thèse pour étudier l’évaluation cognitive réalisée par le leader, au travers de l’activation des caractéristiques utilisées pour catégoriser les suiveurs : les théories implicites du followership (IFTs).L’étude de l’évaluation cognitive du leader se confronte à de nombreuses difficultés, notamment méthodologiques. Ainsi, un travail préliminaire essentiel de ce travail doctoral s’est attaché à traduire et valider en français une échelle d’évaluation des IFTs. La première étude exploratoire utilise des images d’agents virtuels pour explorer plusieurs CNV pouvant activer les IFTs du leader de manière explicite. La deuxième étude propose une analyse de vidéos d’entraînements d’équipes médicales pré-hospitalières pour étudier dynamiquement d’autres CNV intervenant dans la perception du followership. En troisième étude, les CNV observés dans les études précédentes ont été implémentés dans un agent virtuel placé en interaction avec des leaders médicaux. Une tâche de Go / No Go a permis par la suite d’étudier l’activation implicite des IFTs. Enfin, une dernière étude quantitative transversale a tenté d’étudier l’influence de l’évaluation cognitive du suiveur par le leader sur les comportements de leadership.Les résultats de ces études suggèrent que certaines caractéristiques affichées dans les CNV (i.e., dominance, support apporté au leader) peuvent activer les IFTs du leader. De plus, l’évaluation cognitive semble avoir une influence sur les comportements de leadership choisi.Nous discuterons les résultats obtenus et présenterons les contributions scientifiques et pédagogiques de cette thèse. Nous ouvrirons notre réflexion au positionnement épistémologique nécessaire à l’étude des IFTs, ainsi qu’à l’utilisation des agents virtuels dans l’étude de la catégorisation. Le but applicatif de ce travail doctoral est l’implémentation de CNV chez des subordonnés virtuels médicaux dans un serious game permettant la formation de leader médicaux. / Interactions in teamwork and their efficiency are based on a hierarchical system including verticale dyads. This research studies leadership through the social roles link to the hierarchical status of the vertical dyad (i.e., leader for the chief and follower for the subordinate).The influence of follower on leader and leadership is increasingly considered. However, the follower’s behaviors are still unexplored. This multidisciplinary doctoral work (i.e., psychology and affective computing) tries to highlight the influence of follower’s behavior on the leader, and the underlying process of this influence.To do so, we used the follower’s nonverbal behaviors (CNV) as the social clues that can influence the leader and his behaviors. The principles of social cognition are applied in this thesis to study the cognitive evaluation made by the leader. We worked on the activation of the caracteristics used to categorize followers, the implicit followership theories (IFTs).The study of the cognitive evaluation made by the leader comes we multiple issues, including methological ones. Thus, a preliminary work of translation and validation of a scale evaluating the IFTs was realized. The first exploratory study used images of virtual agents displaying CNV that could explicitly activate leaders’ IFTs. We complete these result through an analysis of a corpus of videos filming medical teams’ training. This study allowed us to observe dynamic CNV that could influence the perception of followership. In a third study, we implemented the CNV of the two previous studies in a virtual agent. Medical leaders had to interact with it, and implicit activation of IFTs was collected using a Go / No Go protocol. Finally, using a quantitative approach, we studied the influence of cognitive evaluation of the leader on his leadership behaviors.Results suggest that some caracteristics displayed in CNV (i.e., dominance and support) can activate the leaders’ IFTs. Moreover, the cognitive evaluation of the leader seems to influence his leadership behaviors.We will discuss the results and explain the scientifical and pédagogical contributions of this thesis. We will analyse our problematic through multiple angles, including the epistemologycal point of view allowing the study of IFTs, and the use of virtual agents in the research field of categorization process. The practical application of the doctoral work is the implementation of CNV in virtual subordinates for the deployement of a serious game for medical team leader.

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第19110号 / 情博第556号 / 新制||情||98(附属図書館) / 32061 / 京都大学大学院情報学研究科知能情報学専攻 / (主査)教授 西田 豊明, 教授 乾 敏郎, 教授 河原 達也 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DGAM

Situationen i dagens omvärld har tillsammans med hög inflation och efterdyningar av en pandemi påverkat fastighetsbranschen på många sätt. Ökade elpriser har gjort att drift- och underhållskostnader för fastigheter har ökat markant. För att minska energikostnaderna kan kommersiella fastighetsbolag implementera energieffektiviserande åtgärder. Det kan dock uppstå problematik mellan hyresgästen och fastighetsägaren när det kommer till delade incitament.   Syftet med studien är att undersöka kommersiella fastighetsägares incitament att energieffektivisera befintliga fastigheter med befintliga hyresgäster. Syftet är även att undersöka hur fastighetsägaren arbetar tillsammans med hyresgästen vid energieffektiviserande åtgärder. Energieffektiviserande åtgärder avser i studiens fall investeringar som sänker energikostnaderna. För att uppnå studiens syfte har en kvalitativ metod tillsammans med semistrukturerade intervjuer tillämpats. Intervjustudien genomfördes med sju väl etablerade kommersiella fastighetsbolag. Studiens resultat visar att i de fall där fastighetsägaren upplåter lokaler med kallhyra finns det inga incitament att energieffektivisera. Däremot finns det incitament för fastighetsägare som upplåter lokaler med varmhyra att göra energieffektiviserande åtgärder i befintliga byggnader. Detta incitament stärks med tanke på dagens höga elpriser där fastighetsägare gjort en vinst på investeringen tidigare än vad de skulle ha gjort om elpriserna legat kvar på tidigare nivåer.  Studien visar även att fastighetsägare förhandlar och tillsammans med hyresgästerna kommer fram till lösningar kring energieffektiviserande åtgärder. Ett alternativ kan vara att parterna delar på investeringen. / The situation in today's world, together with high inflation and the aftermath of a pandemic, has affected the property industry in many ways. Increased electricity prices have meant that operating and maintenance costs for properties have increased significantly. In order to overcome energy costs, commercial real estate companies can implement energy efficiency measures. However, problems can arise between the tenant and the property owner when it comes to shared incentives.  The purpose of the study is to investigate the commercial property owner's incentives to make existing properties with existing tenants more energy efficient. The purpose is also to investigate how the property owner works together with the tenant in energy efficiency measures. In the case of the study, energy efficiency measures refer to investments that lower energy costs. In order to achieve the purpose of the study, a qualitative method together with semi-structured interviews has been applied. The interview study was conducted with seven well-established commercial real estate companies. The results of the study show that in cases where the property owner leases premises with cold rent, there is no incentive to improve energy efficiency. On the other hand, there are incentives for property owners, who leases premises with varm rents to make energy efficiency measures in existing buildings. This incentive is also strengthened in view of today's high electricity prices, where property owners have made a profit on the investment earlier than they could have if electricity prices had remained at previous levels. The study also shows that property owners negotiate and together with the tenants reaches solutions regarding energy efficiency measures. An alternative could be for the parties to share the investment.

Throughout today’s society, increasingly more areas are being automated. Grocery stores however have been the same for years. Only recently, self-checkout counters and online shopping have been utilised in this business area. This thesis aims to take it to the next step by introducing automated grocery stores using a multi-agent system. Orders will be given to the system, and on a small area, multiple agents will pick the products in a time-efficient way and deliver them to the customer. This can both increase the throughput but also decrease the food waste and energy consumption of grocery stores. This thesis investigates already existing solutions for the multi-agent pickup and delivery problem. It extends these to the important case of batched tasks in order to improve the customer experience. Batches of tasks represent shopping carts, where fast completion of whole batches gives greater customer satisfaction. This notion is not mentioned in related work, where completion of single tasks is the main goal. Because of this, the existing solution does not accommodate the need of batches or the importance of completing whole batches fast and in somewhat linear order. For this purpose, a new metric called batch ordering weighted error (BOWE) was created that takes these factors into consideration. Using BOWE, one existing algorithm has been extended into prioritizing completing whole batches and is now called B-PIBT. This new algorithm has significantly improved BOWE and even batch service time for the algorithm in key cases and is now superior in comparison to the other state-of-the-art algorithms.

Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

Cooperative multi-agent systems (MAS) are finding applications in a wide variety of domains, including sensor networks, robotics, distributed control, collaborative decision support systems, and data mining. A cooperative MAS consists of a group of autonomous agents that interact with one another in order to optimize a global performance measure. A central challenge in cooperative MAS research is to design distributed coordination policies. Designing optimal distributed coordination policies offline is usually not feasible for large-scale complex multi-agent systems, where 10s to 1000s of agents are involved, there is limited communication bandwidth and communication delay between agents, agents have only limited partial views of the whole system, etc. This infeasibility is either due to a prohibitive cost to build an accurate decision model, or a dynamically evolving environment, or the intractable computation complexity. This thesis develops a multi-agent reinforcement learning paradigm to allow agents to effectively learn and adapt coordination policies in complex cooperative domains without explicitly building the complete decision models. With multi-agent reinforcement learning (MARL), agents explore the environment through trial and error, adapt their behaviors to the dynamics of the uncertain and evolving environment, and improve their performance through experiences. To achieve the scalability of MARL and ensure the global performance, the MARL paradigm developed in this thesis restricts the learning of each agent to using information locally observed or received from local interactions with a limited number of agents (i.e., neighbors) in the system and exploits non-local interaction information to coordinate the learning processes of agents. This thesis develops new MARL algorithms for agents to learn effectively with limited observations in multi-agent settings and introduces a low-overhead supervisory control framework to collect and integrate non-local information into the learning process of agents to coordinate their learning. More specifically, the contributions of already completed aspects of this thesis are as follows: Multi-Agent Learning with Policy Prediction: This thesis introduces the concept of policy prediction and augments the basic gradient-based learning algorithm to achieve two properties: best-response learning and convergence. The convergence property of multi-agent learning with policy prediction is proven for a class of static games under the assumption of full observability. MARL Algorithm with Limited Observability: This thesis develops PGA-APP, a practical multi-agent learning algorithm that extends Q-learning to learn stochastic policies. PGA-APP combines the policy gradient technique with the idea of policy prediction. It allows an agent to learn effectively with limited observability in complex domains in presence of other learning agents. The empirical results demonstrate that PGA-APP outperforms state-of-the-art MARL techniques in both benchmark games. MARL Application in Cloud Computing: This thesis illustrates how MARL can be applied to optimizing online distributed resource allocation in cloud computing. Empirical results show that the MARL approach performs reasonably well, compared to an optimal solution, and better than a centralized myopic allocation approach in some cases. A General Paradigm for Coordinating MARL: This thesis presents a multi-level supervisory control framework to coordinate and guide the agents' learning process. This framework exploits non-local information and introduces a more global view to coordinate the learning process of individual agents without incurring significant overhead and exploding their policy space. Empirical results demonstrate that this coordination significantly improves the speed, quality and likelihood of MARL convergence in large-scale, complex cooperative multi-agent systems. An Agent Interaction Model: This thesis proposes a new general agent interaction model. This interaction model formalizes a type of interactions among agents, called {\em joint-even-driven} interactions, and define a measure for capturing the strength of such interactions. Formal analysis reveals the relationship between interactions between agents and the performance of individual agents and the whole system. Self-Organization for Nearly-Decomposable Hierarchy: This thesis develops a distributed self-organization approach, based on the agent interaction model, that dynamically form a nearly decomposable hierarchy for large-scale multi-agent systems. This self-organization approach is integrated into supervisory control framework to automatically evolving supervisory organizations to better coordinating MARL during the learning process. Empirically results show that dynamically evolving supervisory organizations can perform better than static ones. Automating Coordination for Multi-Agent Learning: We tailor our supervision framework for coordinating MARL in ND-POMDPs. By exploiting structured interaction in ND-POMDPs, this tailored approach distributes the learning of the global joint policy among supervisors and employs DCOP techniques to automatically coordinate distributed learning to ensure the global learning performance. We prove that this approach can learn a globally optimal policy for ND-POMDPs with a property called groupwise observability.