Doctor of Philosophy / Department of Industrial & Manufacturing Systems Engineering / Chih-Hang Wu / The topic of infectious disease epidemics has recently attracted substantial attentions in research communities and it has been shown that the changes of human behaviors have significant impacts on the dynamics of disease transmission. However, the study and understanding of human reactions into spread of infectious disease are still in the very beginning phase and how human behaviors change during the spread of infectious disease has not been systematically investigated. Moreover, the study of human behaviors includes not only various enforced measures by public authorities such as school closure, quarantine, vaccination, etc, but also the spontaneous self-protective actions which are triggered by risk perception and fear of diseases. Hence, the goal of this research is to study the impacts of human behaviors to the epidemic from these two perspectives: spontaneous behavioral changes and public intervention strategies. For the sake of studying spontaneous changes of human behaviors, this research first time applied evolutionary spatial game into the study of human reactions to the spread of infectious disease. This method integrated contact structures and epidemics information into the individuals’ decision processes, by adding two different types of information into the payoff functions: the local information and global information. The new method would not only advance the field of game theory, but also the field of epidemiology. In addition, this method was also applied to a classic compartmental dynamic system which is a widely used model for studying the disease transmission. With extensive numerical studies, the results first proved the consistency of two models for the sake of validating the effectiveness of the spatial evolutionary game. Then the impacts of changes of human behaviors to the dynamics of disease transmission and how information impacts human behaviors were discussed temporally and spatially. In addition to the spontaneous behavioral changes, the corresponding intervention strategies by policy-makers played the key role in process of mitigating the spread of infectious disease. For the purpose of minimizing the total lost, including the social costs and number of infected individuals, the intervention strategies should be optimized. Sensitivity analysis, stability analysis, bifurcation analysis, and optimal control methods are possible tools to understand the effects of different combination of intervention strategies or even find an appropriate policy to mitigate the disease transmission. One zoonotic disease, named Zoonotic Visceral Leishmaniasis (ZVL), was studied by adopting different methods and assumptions. Particularly, a special case, backward bifurcation, was discussed for the transmission of ZVL. Last but not least, the methodology and modeling framework used in this dissertation can be expanded to other disease situations and intervention applications, and have a broad impact to the research area related to mathematical modeling, epidemiology, decision-making processes, and industrial engineering. The further studies can combine the changes of human behaviors and intervention strategies by policy-makers so as to seek an optimal information dissemination to minimize the social costs and the number of infected individuals. If successful, this research should aid policy-makers by improving communication between them and the public, by directing educational efforts, and by predicting public response to infectious diseases and new risk management strategies (regulations, vaccination, quarantine, etc.).

abstract: Emerging trends in cyber system security breaches in critical cloud infrastructures show that attackers have abundant resources (human and computing power), expertise and support of large organizations and possible foreign governments. In order to greatly improve the protection of critical cloud infrastructures, incorporation of human behavior is needed to predict potential security breaches in critical cloud infrastructures. To achieve such prediction, it is envisioned to develop a probabilistic modeling approach with the capability of accurately capturing system-wide causal relationship among the observed operational behaviors in the critical cloud infrastructure and accurately capturing probabilistic human (users’) behaviors on subsystems as the subsystems are directly interacting with humans. In our conceptual approach, the system-wide causal relationship can be captured by the Bayesian network, and the probabilistic human behavior in the subsystems can be captured by the Markov Decision Processes. The interactions between the dynamically changing state graphs of Markov Decision Processes and the dynamic causal relationships in Bayesian network are key components in such probabilistic modelling applications. In this thesis, two techniques are presented for supporting the above vision to prediction of potential security breaches in critical cloud infrastructures. The first technique is for evaluation of the conformance of the Bayesian network with the multiple MDPs. The second technique is to evaluate the dynamically changing Bayesian network structure for conformance with the rules of the Bayesian network using a graph checker algorithm. A case study and its simulation are presented to show how the two techniques support the specific parts in our conceptual approach to predicting system-wide security breaches in critical cloud infrastructures. / Dissertation/Thesis / Masters Thesis Computer Science 2015

abstract: Cyber systems, including IoT (Internet of Things), are increasingly being used ubiquitously to vastly improve the efficiency and reduce the cost of critical application areas, such as finance, transportation, defense, and healthcare. Over the past two decades, computing efficiency and hardware cost have dramatically been improved. These improvements have made cyber systems omnipotent, and control many aspects of human lives. Emerging trends in successful cyber system breaches have shown increasing sophistication in attacks and that attackers are no longer limited by resources, including human and computing power. Most existing cyber defense systems for IoT systems have two major issues: (1) they do not incorporate human user behavior(s) and preferences in their approaches, and (2) they do not continuously learn from dynamic environment and effectively adapt to thwart sophisticated cyber-attacks. Consequently, the security solutions generated may not be usable or implementable by the user(s) thereby drastically reducing the effectiveness of these security solutions. In order to address these major issues, a comprehensive approach to securing ubiquitous smart devices in IoT environment by incorporating probabilistic human user behavioral inputs is presented. The approach will include techniques to (1) protect the controller device(s) [smart phone or tablet] by continuously learning and authenticating the legitimate user based on the touch screen finger gestures in the background, without requiring users’ to provide their finger gesture inputs intentionally for training purposes, and (2) efficiently configure IoT devices through controller device(s), in conformance with the probabilistic human user behavior(s) and preferences, to effectively adapt IoT devices to the changing environment. The effectiveness of the approach will be demonstrated with experiments that are based on collected user behavioral data and simulations. / Dissertation/Thesis / Doctoral Dissertation Computer Science 2016

Cette thèse s’intéresse à la simulation de comportements humains en milieu urbain. Elle se focalise tout particulièrement sur la crédibilité des comportements des agents, telle que jugée par un observateur externe. Pour cela, nos agents sont dotés de capacités d’anticipation, qui leur permet de gagner en efficacité en étant capables d’utiliser des prédictions concernant leur futur proche. Par ailleurs, l’architecture d’agents utilisée se place dans le paradigme des architectures hybrides, mais fait preuve d’une grande innovation par rapport à l’état de l’art en raison de son organisation résolument modulaire, permettant une grande généricité du modèle. En effet, l’architecture fonctionne avec un système de modules de haut-niveau, étant vu comme des boîtes noires par le reste de l’architecture. Leur nombre et leur modélisation interne sont donc entièrement libres. Cette généricité est très intéressante, puisque le domaine de la simulation urbaine touche de très nombreux domaines applicatifs (urbanisme, transport, jeu vidéo, sécurité, etc.) ayant des contraintes différentes. Un modèle d’agent générique permet de s’adapter à l’application désirée. Cette généricité pose néanmoins un problème lié à l’intégration et au traitement d’un grand nombre de comportements hétérogènes au sein d’un même processus décisionnel, problème que nous traitons grâce à un mécanisme de composition de comportements. Pour finir, notre architecture permet d’assurer le passage à l’échelle en servant de plusieurs niveaux de détail dans la modélisation des agents. / This thesis deals about human behaviors simulation in an urban context. We focus on the behavior believability (as judged as external observers). That is why our agent have some anticipatory skills, which allow them to use predictions about their near future. Our architecture is a hybrid one, which is very innovative because of its functioning with « high-level modules », which are seen as black-box from the rest of the architecture. Their number and intern modeling are completely free. This makes our architecture very modular and generic, and it is important because the urban simulation domain has many different applications (urbanism, video games, security, etc.), with different constraints. However, this genericity brings another problem, which is the integration of several heterogeneous behavior into the same decisional process. This issue is addressed thanks to a behavior composition mechanism. To conclude, we ensure the scaling up of our architecture with the creation of several levels of detail in the agents modeling.

Les technologies d’assistance à la vie autonome est aujourd'hui nécessaire pour soutenir les personnes ayant des besoins spécifiques dans leurs activités de la vie quotidienne, mais leurs développements demeure limités malgré les enjeux liés à l’accompagnement des personnes âgées et dépendantes. Par ailleurs, l'élaboration de plateformes technologiques durant la dernière décennie s'est principalement concentrée sur la dimension technologique, en négligeant l'impact des facteurs humains et des besoins sociaux. Les nouvelles technologies, telles que le cloud et l’Internet des objets (IoT) pourraient apporter de nouvelles capacités dans ce domaine de recherche permettant aux systèmes de traiter les activités humaines selon des modèles orientés vers l'usage (ie. la fragilité) dans une approche non invasive.Cette thèse se propose d'envisager un nouveau paradigme dans les technologies d'assistance pour le vieillissement et le bien-être en introduisant (i) des métriques de la fragilité humaine et (ii) une dimension urbaine dans un cadre d'assistance ambiant (extension de l'espace de vie de l'intérieur vers l'extérieur). Elle propose une plateforme basée sur l’Informatique dans le cloud (cloud computing) pour une communication transparente avec les objets connectés, permettant au système intégré de calculer et de modéliser différents niveaux de fragilité humaine. Cette thèse propose d'utiliser des données hétérogènes en temps réel fournies par différents types de sources (capteurs intérieurs et extérieurs), ainsi que des données de référence, collectées sur un serveur de cloud de raisonnement central. La plateforme stocke les données brutes et les traite à travers un moteur de raisonnement hybride combinant à la fois l'approche basée sur les données (apprentissage automatique), et l'approche basée sur la connaissance (raisonnement sémantique) pour (i) déduire les activités de la vie quotidienne, (ii) détecter le changement du comportement humain, et enfin (iii) calibrer les valeurs de fragilité humaine. Les valeurs de fragilité peuvent permettre au système de détecter automatiquement tout changement de comportement, ou toute situation anormale, qui pourrait entraîner un risque à la maison ou à l'extérieur. L'ambition à long terme est de détecter et d'intervenir pour éviter un risque avant même qu'un médecin ne le détecte lors d'une consultation. L'objectif ultime est de promouvoir le paradigme de la prévention pour la santé et du bien-être.Cette thèse vise à concevoir et développer une plateforme intégrée, personnalisée, basée sur le cloud, capable de communiquer avec des capteurs intérieurs non invasifs (par ex. mouvement, contact, fibre optique) et à l'extérieur (par ex. BLE Beacons, smartphone, bracelet..). La plateforme développée comprend également un classificateur de mobilité du comportement humain qui utilise les capteurs internes du Smart Phone pour calibrer le type de mouvement effectué par l'individu (p. ex. marche, vélo, tram, bus, et voiture). Les données recueillies dans ce contexte servent à construire un modèle multidimensionnel de fragilité basé sur plusieurs éléments standardisés de fragilité, à partir d'une littérature abondante et d'un examen approfondi d’autres plateformes. La plateforme et les modèles associés ont été évalués dans des conditions réelles de vie impliquant les utilisateurs et les aidants par le biais de différents sites pilotes à Singapour et en France. Les données obtenues ont été analysées et publiées dans de nombreuses conférences et revues internationales.La plateforme développée est actuellement déployée en situation écologique dans 24 habitats individuels. Cela comprend cinq chambres en EHPAD, et neuf maisons sont situées en France, en collaboration avec une maison de retraite (Argentan-Normandie) et à Montpellier en collaboration avec Montpellier Métropole. Entre autre dix appartements privés sont situés à Singapour en collaboration avec un Senior Activity Center. / Ambient Assisted Living is nowadays necessary to support people with special needs in performing their activities of daily living, but it remains unaltered in front of the necessity to accompany aging and dependent people in their outdoors activities. Moreover, the development of multiple frameworks during the last decade has mainly focused on the engineering dimension neglecting impact of human factors and social needs in the design process. New technologies, such as cloud computing and Internet of Things (IoT) could bring new capabilities to this field of research allowing systems to process human condition following usage oriented models (e.g. frailty) in a non-invasive approach. This thesis proposes to consider a new paradigm in assistive technologies for aging and wellbeing by introducing (i) human frailty metrics, and (ii) urban dimension in an ambient assistive framework (extending the living space from indoors to outdoors). It proposes a cloud-based framework for seamless communication with connected objects, allowing the integrated system to compute and to model different levels of human frailty based on several frailty standardized items, and leveraged from an extensive literature and frameworks reviews.This thesis aims at designing and developing an integrated cloud-based framework, which would be able to communicate with heterogeneous real-time non-invasive indoor sensors (e.g. motion, contact, fiber optic) and outdoors (e.g. BLE Beacons, smartphone). The framework stores the raw data and processes it through a designed hybrid reasoning engine combining both approaches, data driven (machine learning), and knowledge driven (semantic reasoning) algorithms, to (i) infer the activities of the daily living (ADL), (ii) detect changes of human behavior, and ultimately (iii) calibrate human frailty values. It also includes a human behavior mobility classifier that uses the inner smartphone sensors to classify the type of movement performed by the individual (e.g. Walk, Cycling, MRT, Bus, Car). The frailty values might allow the system to automatically detect any change of behaviors, or abnormal situations, which might lead to a risk at home or outside.The proposed models and framework have been developed in close collaboration with IPAL and LIRMM research teams. They also have been assessed in real conditions involving end-users and caregivers through different pilots sites in Singapore and in France. Nowadays, the proposed framework, is currently deployed in a real world deployment in 24 individual homes. 14 spaces are located in France (5 privates rooms in nursing home and 9 private houses) in collaboration with a nursing home (Argentan-Normandie and Montpellier). 10 individual homes are located in Singapore in collaboration with a Senior Activity Center (non-profit organization).The long-term ambition is to detect and intervene to avoid a risk even before a medical doctor detects it during a consultation. The ultimate goal is to promote prevention paradigm for health and wellbeing. The obtained data has been analyzed and published in multiple international conferences and journals.