In order to explore further the capabilities of parallel computing architectures such as grids, clusters, multi-processors and more recently, clouds and multi-cores, an easy-to-use parallel language is an important challenging issue. From the programmer's point of view, OpenMP is very easy to use with its ability to support incremental parallelization, features for dynamically setting the number of threads and scheduling strategies. However, as initially designed for shared memory systems, OpenMP is usually limited on distributed memory systems to intra-nodes' computations. Many attempts have tried to port OpenMP on distributed systems. The most emerged approaches mainly focus on exploiting the capabilities of a special network architecture and therefore cannot provide an open solution. Others are based on an already available software solution such as DMS, MPI or Global Array and, as a consequence, they meet difficulties to become a fully-compliant and high-performance implementation of OpenMP. As yet another attempt to built an OpenMP compliant implementation for distributed memory systems, CAPE − which stands for Checkpointing Aide Parallel Execution − has been developed which with the following idea: when reaching a parallel section, the master thread is dumped and its image is sent to slaves; then, each slave executes a different thread; at the end of the parallel section, slave threads extract and return to the master thread the list of all modifications that has been locally performed; the master includes these modifications and resumes its execution. In order to prove the feasibility of this paradigm, the first version of CAPE was implemented using complete checkpoints. However, preliminary analysis showed that the large amount of data transferred between threads and the extraction of the list of modifications from complete checkpoints lead to weak performance. Furthermore, this version was restricted to parallel problems satisfying the Bernstein's conditions, i.e. it did not solve the requirements of shared data. This thesis aims at presenting the approaches we proposed to improve CAPE' performance and to overcome the restrictions on shared data. First, we developed DICKPT which stands for Discontinuous Incremental Checkpointing, an incremental checkpointing technique that supports the ability to save incremental checkpoints discontinuously during the execution of a process. Based on the DICKPT, the execution speed of the new version of CAPE was significantly increased. For example, the time to compute a large matrix-matrix product on a desktop cluster has become very similar to the execution time of the same optimized MPI program. Moreover, the speedup associated with this new version for various number of threads is quite linear for different problem sizes. In the side of shared data, we proposed UHLRC, which stands for Updated Home-based Lazy Release Consistency, a modified version of the Home-based Lazy Release Consistency (HLRC) memory model, to make it more appropriate to the characteristics of CAPE. Prototypes and algorithms to implement the synchronization and OpenMP data-sharing clauses and directives are also specified. These two works ensures the ability for CAPE to respect shared-data behavior

Draft and proof of an algorithm computing incremental changes within a labeled layouted concept lattice upon insertion or removal of an attribute column in the underlying formal context. Furthermore some implementational details and mathematical background knowledge are presented. / Entwurf und Beweis eines Algorithmus zur Berechnung inkrementeller Änderungen in einem beschrifteten dargestellten Begriffsverband beim Einfügen oder Entfernen einer Merkmalsspalte im zugrundeliegenden formalen Kontext. Weiterhin sind einige Details zur Implementation sowie zum mathematischen Hintergrundwissen dargestellt.

Precast prestressed hollow core units are commonly used in the construction of the flooring system in precast buildings. These units without transverse reinforcement bars are designed to resist seismic loading as replacement for fixed-base precast wall panels in the construction of warehouse buildings. Thus, this research seeks to investigate the seismic performance of the units constructed as a subassemblage (single wall) subjected to biaxial loading and as a superassemblage (multi-panel) subjected to quasi-static lateral loading. A design procedure for warehouse building using precast hollow core walls under Damage Avoidance Design (DAD) is proposed. In addition, a risk assessment under Performance-Based Earthquake Engineering (PBEE) is evaluated using the latest computational tool known as Incremental Dynamic Analysis (IDA). A comparative risk assessment between precast hollow core walls and fixed-base monolithic precast wall panels is also performed. Experimental results demonstrate that rocking precast hollow core walls with steelarmouring do not suffer any non-structural damage up to 2.0% drift and minor structural damage at 4.0% drift. Results revealed that the wall with unbonded fuse-bars and 50% initial prestressing of unbonded tendons performed the best compared with other types of energy dissipators. Furthermore, 12mm diameter of fuse-bar is recommended as there is no uplifting of the foundation beam during ground shaking. Hence, this type of energy dissipator is used for the construction of seismic wall panels in warehouse buildings. One of the significant findings is that the capacity reduction factor (Ø ) which relates to global uncertainty of seismic performance is approximately equal to 0.6. This value can be used to estimate the 90th percentile of the structures without performing IDA. Therefore, the structural engineers are only required to compute Rapid-IDA curve along with the proposed design procedure.

Incremental Dynamic Analysis (IDA) procedures are advanced and then applied to a quantitative risk assessment for bridge structures. This is achieved by combining IDA with site-dependent hazard-recurrence relations and damage outcomes. The IDA procedure is also developed as a way to select a critical earthquake motion record for a one-off destructive experiment. Three prototype bridge substructures are designed according to the loading and detailing requirements of New Zealand, Japan and Caltrans codes. From these designs 30 percent reduced scale specimens are constructed as part of an experimental investigation. The Pseudodynamic test is then to control on three specimens using the identified critical earthquake records. The results are presented in a probabilistic riskbased format. The differences in the seismic performance of the three different countries' design codes are examined. Each of these current seismic design codes strive for ductile behaviour of bridge substructures. Seismic response is expected to be resulting damage on structures, which may threaten post-earthquake serviceability. To overcome this major performance shortcoming, the seismic behaviour under bi-directional lateral loading is investigated for a bridge pier designed and constructed in accordance with Damage Avoidance principles. Due to the presence of steel armoured rocking interface at the base, it is demonstrated that damage can be avoided, but due to the lack of hysteresis it is necessary to add some supplemental damping. Experimental results of the armoured rocking pier under bi-directional loading are compared with a companion ductile design specimen.

This dissertation explores the intersection of information retrieval and universal source coding techniques and studies an optimal multidimensional source representation from an information theoretic point of view. Previous research on information retrieval particularly focus on learning probabilistic or deterministic source models based on primarily two different types of source representations, e.g., fixed-shape partitions or uniform regions. We study the limitations of the conventional source representations on capturing the semantics of the given multidimensional source sequences and propose a new type of primitive source representation generated by a universal source coding technique. We propose a multidimensional incremental parsing algorithm extended from the Lempel-Ziv incremental parsing and its three component schemes for multidimensional source coding. The properties of the proposed coding algorithm are exploited under two-dimensional lossless and lossy source coding. By the proposed coding algorithm, a given multidimensional source sequence is parsed into a number of variable-size patches. We call this methodology a parsed representation. Based on the source representation, we propose an information retrieval framework that analyzes a set of source sequences under a linguistic processing technique and implemented content-based image retrieval systems. We examine the relevance of the proposed source representation by comparing it with the conventional representation of visual information. To further extend the proposed framework, we apply a probabilistic linguistic processing technique to modeling the latent aspects of a set of documents. In addition, beyond the symbol-wise pattern matching paradigm employed in the source coding and the image retrieval systems, we devise a robust pattern matching that compares the first- and second-order statistics of source patches. Qualitative and quantitative analysis of the proposed framework justifies the superiority of the proposed information retrieval framework based on the parsed representation. The proposed source representation technique and the information retrieval frameworks encourage future work in exploiting a systematic way of understanding multidimensional sources that parallels a linguistic structure.

Les estimations précises des états de charge (« State of Charge » - SoC) et de santé (« State of Health » - SoH) des batteries au lithium sont un point crucial lors d’une utilisation industrielle de celles-ci. Ces estimations permettent d’améliorer la fiabilité et la robustesse des équipements embarquant ces batteries. Cette thèse CIFRE est consacrée à la recherche d’algorithmes de détermination de l’état de santé de batteries lithium-ion, en particulier de chimie Lithium Fer Phosphate (LFP) et Lithium Manganèse Oxyde (LMO).Les recherches ont été orientées vers des solutions de détermination du SoH directement embarquables dans les calculateurs des véhicules électriques. Des contraintes fortes de coût et de robustesse constituent ainsi le fil directeur des travaux.Or, si la littérature actuelle propose différentes solutions de détermination du SoH, celles embarquées ou embarquables sont encore peu étudiées. Cette thèse présente donc une importante revue bibliographique des différentes méthodes d’estimation du SoH existantes, qu’elles soient embarquables ou non. Le fonctionnement détaillé ainsi que les mécanismes de vieillissement d’une batterie lithium-ion sont également explicités.Une partie majoritaire des travaux est consacrée à l’utilisation de l’analyse incrémentale de la capacité (« Incremental Capacity Analysis » - ICA) en conditions réelles, c’est-à-dire avec les niveaux de courant présents lors d’un profil de mission classique d’un véhicule électrique, avec les mesures disponibles sur un BMS (« Battery Management System ») industriel et avec les contraintes de robustesses associées, notamment une gamme étendue de température de fonctionnement. L’utilisation de l’ICA pour déterminer la capacité résiduelle de la batterie est mise en œuvre de façon totalement innovante et permet d’obtenir une grande robustesse aux variations des conditions d’utilisation de la batterie.Une seconde méthode est, elle, dédiée à la chimie LMO et exploite le fait que le potentiel aux bornes de la batterie soit représentatif de son état de charge. Un compteur coulométrique partiel est ainsi proposé, intégrant une gestion dynamique des bornes d’intégration en fonction de l’état de la batterie.A l’issue des travaux, une méthode complète et précise de détermination du SoH est disponible pour chacune des chimies LFP et LMO. La détermination de la capacité résiduelle de ces deux familles de batteries est ainsi possible à 4 % près. / Accurate lithium-ion battery State of Charge (SoC) and State of Health (SoH) estimations are nowadays a crucial point, especially when considering an industrial use. These estimations enable to improve robustness and reliability of hardware using such batteries. This thesis focuses on researching lithium-ion batteries state of health estimators, in particular considering Lithium Iron Phosphate (LFP) and Lithium Manganese Oxide (LMO) chemistries.Researches have been targeted towards SoH estimators straight embeddable into electric vehicles (EV) computers. Cost and reliability constraints are thus the main guideline for this work.Although existing literature offers various SoH estimators, those who are embedded or embeddable are still little studied. A complete literature review about SoH estimators, embedded or not, is therefore proposed. Lithium-ion batteries detailed operation and ageing mechanisms are also presented.The main part of this work is dedicated to Incremental Capacity Analysis (ICA) use with electric vehicle constraints, such as current levels available with a typical EV mission profile or existing measurements on the Battery Management System (BMS). Incremental Capacity Analysis is implemented in an innovative way and leads to a remaining capacity estimator with a high robustness to conditions of use variations, including an extended temperature range.A second method, dedicated to LMO chemistry, take advantage of the fact that the battery potential is representative of its state of charge. Partial Coulomb counting is thus performed, with a dynamic management of integration limits, depending on the battery state.Outcomes of this work are two complete and accurate SoH estimators, one for each chemistry, leading to a remaining capacity estimation accurate within 4 %.

Les travaux de recherche présentés et discutés dans cette thèse s’intéressent aux grandes collections d’images numériques. Plus particulièrement, nous cherchons à donner à un utilisateur la possibilité d’explorer ces collections d’images, soit dans le but d’en extraire de l’information et de la connaissance, soit de permettre une certaine sérendipité dans l’exploration. Ainsi, cette problématique est abordée du point de vue de l’analyse et l’exploration interactive des données. Nous tirons profit du paradigme de navigation par similarité et visons à respecter simultanément les trois contraintes suivantes : (i) traiter de grandes collections d’images, (ii) traiter des collections dont le nombre d’images ne cesse de croître au cours du temps et (iii) donner des moyens d’explorer interactivement des collections d’images. Pour ce faire, nous proposons d’effectuer une étude conjointe de l’indexation et de la visualisation de grandes collections d’images qui s’agrandissent au cours du temps. / The research work that is presented and discussed in this thesis focuses on large and evergrowing image collections. More specifically, we aim at providing one the possibility to explore such image collections, either to extract some kind of information and knowledge, or to wander in the collections. This thesis addresses this issue from the perspective of Interactive Data Exploration and Analytics. We take advantage of the similarity-based image collection browsing paradigm and aim at meeting simultaneously the three following constraints: (i) handling large image collections, up to millions of images, (ii) handling dynamic image collections, to deal with ever-growing image collections, and (iii) providing interactive means to explore image collections. To do so, we jointly study the indexing and the interactive visualisation of large and ever-growing image collections.

Afin d'approfondir la compréhension des mécanismes de vieillissement des batteries Li-ion, des analyses post-mortem ont été effectuées sur des cellules commerciales Li-ion C/NMC. Ces autopsies ont révélé des dégradations inattendues qui remettent en question les connaissances actuelles sur les mécanismes de vieillissement de ces cellules. Ainsi, il semble que la réaction parasite des dépôts de Li métallique sur l'électrode en graphite, actuellement associée dans la littérature à des charges à basses températures et / ou à courants élevés, aurait diverses origines selon la chimie et les conditions d'utilisation de la batterie. Dans ce travail de thèse, des dépôts locaux de Li métallique ont été observés sur des cellules vieillies en calendaire à haute température. Paradoxalement, dans des conditions de cyclage à basse température, ce dépôt de Li métallique a résulté de la perte de porosité au niveau de l’électrode négative. Par ailleurs, un modèle de vieillissement semi-empirique, prenant compte les pertes en cyclage ainsi que celles causées par la croissance de la SEI et la polymérisation du biphényl, est proposé. Pour finir, une méthode d'identification des modes de dégradation grâce à des mesures de capacité incrémentale a été entreprise, sur la base du décalage des potentiels de chacune des électrodes. / In order to deepen the understanding of the aging mechanisms of Li-ion batteries, post-mortem investigations were performed on C/NMC Li-ion commercial cells. These autopsies revealed unexpected degradations that question current knowledge about the aging mechanisms of these cells. Thus, it appears that the parasitic reaction of Li metal depositions on the graphite electrode, nowadays associated in the literature with charging at low temperature and / or high C-rates, would have various origins depending on the chemistry and conditions of use of the battery. In this thesis work, local Li deposits were observed on cells aged in calendar at high temperatures, due to the apparition of dry areas. Paradoxically, under low temperature cycling conditions, this Li resulted from anode porosity hindrance. Besides, a semi-empirical aging model, taking into account cycling losses as well as those caused by the SEI growth and the biphenyl polymerization, is proposed. Finally, a method of identifying degradation modes using incremental capacity measurements has been undertaken, based on the potential shifts of each of the electrodes.

[pt] Aplicações ubíquas estão inseridas em ambientes inteligentes integrados ao mundo físico e compostos de usuários com diferentes preferências, dispositivos heterogêneos e vários provedores de serviço e conteúdo. Além disso, essas aplicações são especializadas em oferecer serviços e conteúdos em qualquer lugar e momento, auxiliando os usuários em suas atividades diárias sem incomodá-los. Baseado nesse mundo idealizado, o paradigma em qualquer lugar e momento impõe alguns desafios para a comunidade de Engenharia de Software, tais como: heterogeneidade de dispositivos, ambientes distribuídos, mobilidade, satisfação de usuário, adaptação de conteúdo, sensibilidade de contexto, privacidade, personalização, transparência, invisibilidade e constante evolução das tendências tecnológicas. Visando lidar com esses novos desafios tecnológicos, é proposta uma abordagem orientada à reutilização de software para desenvolvimento incremental e sistemático de aplicações ubíquas intencionais. Foram escolhidos dois principais objetivos para conduzir a pesquisa dessa tese: (i) a construção de conjuntos de apoio, orientados à reutilização de software, com base em uma investigação detalhada de aplicações ubíquas e do paradigma de Sistemas Multi-Agentes Intencionais – ou seja, Desenvolvimento para Reutilização; e (ii) o desenvolvimento incremental e sistemático de aplicações ubíquas, dirigidas por Sistemas Multi-Agentes Intencionais, com base na abordagem orientada à reutilização de software – ou seja, Desenvolvimento com Reutilização. Algumas contribuições do nosso trabalho são: (i) uma arquitetura orientada à reutilização de software e baseada nos conjuntos de apoio – i.e. blocos de construção principalmente compostos de modelos conceituais, frameworks, padrões e bibliotecas – obtidos a partir da Engenharia de Domínio das Aplicações Ubíquas; (ii) uma Engenharia de Aplicações Ubíquas orientada à reutilização de software visando o desenvolvimento incremental e sistemático de aplicações ubíquas com base nos blocos de construção propostos; (iii) um modelo de raciocínio focado em regras condicionais de lógica nebulosa e no modelo Crença-Desejo-Intenção para melhorar a capacidade cognitiva dos agentes; (iv) um mecanismo específico, baseado em agentes intencionais, para lidar com questões de privacidade, balanceando privacidade e personalização bem como transparência e invisibilidade; (v) um catálogo que graficamente apresenta os principais requisitos não-funcionais ubíquos, as interdependências entre eles e formas de se operacionalizá-los com base na combinação de tecnologias tradicionais e emergentes; (vi) ontologias para permitir a construção dinâmica de interfaces e melhorar a comunicação e inter-operabilidade dos agentes de software; e (vii) um modelo de banco de dados dinâmico para carregar e recuperar os perfis ubíquos (ex. perfis de usuário, dispositivo, rede e contrato), melhorando o gerenciamento de dados em tempo de execução. A abordagem proposta foi avaliada desenvolvendo diferentes aplicações ubíquas (ex. aplicações ubíquas de comércio eletrônico e de clínica odontológica). / [en] Ubiquitous applications are embedded in intelligent environments integrated into the physical world and composed of users with different preferences, heterogeneous devices and several content and service providers. Moreover, they focus on offering services and contents anywhere and at any time by assisting the users in their daily activities without disturbing them. Based on this idealized world, the anywhere and at any time paradigm poses some challenges for the Software Engineering community, such as: device heterogeneity, distributed environments, mobility, user satisfaction, content adaptability, context awareness, privacy, personalization, transparency, invisibility and constant evolution of technological trends. In order to deal with these new technological challenges, we propose a Reuse-Oriented Approach for Incremental and Systematic Development of Intentional Ubiquitous Applications. We have chosen two main goals that drive our research in this thesis: (i) the construction of reuse-oriented support sets based on an extensive investigation of ubiquitous applications and the Intentional-Multi-Agent Systems paradigm – i.e. Development for Reuse; and (ii) the incremental and systematic development of Intentional-Multi-Agent-Systems-driven ubiquitous applications based on the reuse-oriented approach – i.e. Development with Reuse. Some contributions of our work are: (i) a reuse-oriented architecture centered on support sets – i.e. building blocks mainly composed of conceptual models, frameworks, patterns and libraries – obtained from the Domain Engineering of Ubiquitous Applications; (ii) a reuse-oriented Ubiquitous Application Engineering for incremental and systematic development of intentional ubiquitous applications centered on the proposed building blocks; (iii) a reasoning engine focused on fuzzy conditional rules and the Belief-Desire-Intention model to improve the agents cognitive capacity; (iv) a specific mechanism based on intentional agents to deal with privacy issues by balancing privacy and personalization as well as transparency and invisibility; (v) a catalogue that graphically presents the main ubiquitous non-functionalrequirements, their interdependencies and ways to operationalize them based on the combination of traditional and emergent technologies; (vi) ontologies to allow the dynamic construction of interfaces and to improve the communication and inter-operability of software agents; and (vii) a dynamic database model to store and retrieve the ubiquitous profiles (e.g. user, device, network and contract profiles) by improving the data management on the fly. The proposed approach was evaluated by developing different ubiquitous applications (e.g. e-commerce and dental clinic ubiquitous applications).

Este trabalho propõe uma nova abordagem conexionista, chamada de IGMN (do inglês Incremental Gaussian Mixture Network), para aproximação incremental de funções e tarefas de tempo real. Ela é inspirada em recentes teorias do cérebro, especialmente o MPF (do inglês Memory-Prediction Framework) e a Inteligência Artificial Construtivista, que fazem com que o modelo proposto possua características especiais que não estão presentes na maioria dos modelos de redes neurais existentes. Além disso, IGMN é baseado em sólidos princípios estatísticos (modelos de mistura gaussianos) e assintoticamente converge para a superfície de regressão ótima a medida que os dados de treinamento chegam. As principais vantagens do IGMN em relação a outros modelos de redes neurais são: (i) IGMN aprende instantaneamente analisando cada padrão de treinamento apenas uma vez (cada dado pode ser imediatamente utilizado e descartado); (ii) o modelo proposto produz estimativas razoáveis baseado em poucos dados de treinamento; (iii) IGMN aprende de forma contínua e perpétua a medida que novos dados de treinamento chegam (não existem fases separadas de treinamento e utilização); (iv) o modelo proposto resolve o dilema da estabilidade-plasticidade e não sofre de interferência catastrófica; (v) a topologia da rede neural é definida automaticamente e de forma incremental (novas unidades são adicionadas sempre que necessário); (vi) IGMN não é sensível às condições de inicialização (de fato IGMN não utiliza nenhuma decisão e/ou inicialização aleatória); (vii) a mesma rede neural IGMN pode ser utilizada em problemas diretos e inversos (o fluxo de informações é bidirecional) mesmo em regiões onde a função alvo tem múltiplas soluções; e (viii) IGMN fornece o nível de confiança de suas estimativas. Outra contribuição relevante desta tese é o uso do IGMN em importantes tarefas nas áreas de robótica e aprendizado de máquina, como por exemplo a identificação de modelos, a formação incremental de conceitos, o aprendizado por reforço, o mapeamento robótico e previsão de séries temporais. De fato, o poder de representação e a eficiência e do modelo proposto permitem expandir o conjunto de tarefas nas quais as redes neurais podem ser utilizadas, abrindo assim novas direções nos quais importantes contribuições do estado da arte podem ser feitas. Através de diversos experimentos, realizados utilizando o modelo proposto, é demonstrado que o IGMN é bastante robusto ao problema de overfitting, não requer um ajuste fino dos parâmetros de configuração e possui uma boa performance computacional que permite o seu uso em aplicações de controle em tempo real. Portanto pode-se afirmar que o IGMN é uma ferramenta de aprendizado de máquina bastante útil em tarefas de aprendizado incremental de funções e predição em tempo real. / This work proposes IGMN (standing for Incremental Gaussian Mixture Network), a new connectionist approach for incremental function approximation and real time tasks. It is inspired on recent theories about the brain, specially the Memory-Prediction Framework and the Constructivist Artificial Intelligence, which endows it with some unique features that are not present in most ANN models such as MLP, RBF and GRNN. Moreover, IGMN is based on strong statistical principles (Gaussian mixture models) and asymptotically converges to the optimal regression surface as more training data arrive. The main advantages of IGMN over other ANN models are: (i) IGMN learns incrementally using a single scan over the training data (each training pattern can be immediately used and discarded); (ii) it can produce reasonable estimates based on few training data; (iii) the learning process can proceed perpetually as new training data arrive (there is no separate phases for leaning and recalling); (iv) IGMN can handle the stability-plasticity dilemma and does not suffer from catastrophic interference; (v) the neural network topology is defined automatically and incrementally (new units added whenever is necessary); (vi) IGMN is not sensible to initialization conditions (in fact there is no random initialization/ decision in IGMN); (vii) the same neural network can be used to solve both forward and inverse problems (the information flow is bidirectional) even in regions where the target data are multi-valued; and (viii) IGMN can provide the confidence levels of its estimates. Another relevant contribution of this thesis is the use of IGMN in some important state-of-the-art machine learning and robotic tasks such as model identification, incremental concept formation, reinforcement learning, robotic mapping and time series prediction. In fact, the efficiency of IGMN and its representational power expand the set of potential tasks in which the neural networks can be applied, thus opening new research directions in which important contributions can be made. Through several experiments using the proposed model it is demonstrated that IGMN is also robust to overfitting, does not require fine-tunning of its configuration parameters and has a very good computational performance, thus allowing its use in real time control applications. Therefore, IGMN is a very useful machine learning tool for incremental function approximation and on-line prediction.