[pt] À medida que os modelos de aprendizado de máquina penetram áreas críticas como medicina, sistema de justiça criminal e mercados financeiros, sua opacidade, que impede que as pessoas interpretem a maioria deles, se tornou um problema a ser resolvido. Neste trabalho, apresentamos uma nova taxonomia para classificar qualquer método, abordagem ou estratégia para lidar com o problema da interpretabilidade de modelos de aprendizado de máquina. A taxonomia proposta que preenche uma lacuna existente nas estruturas de taxonomia atuais em relação à percepção subjetiva de diferentes intérpretes sobre um mesmo modelo. Para avaliar a taxonomia proposta, classificamos as contribuições de artigos científicos relevantes da área. / [en] As machine learning models penetrate critical areas like medicine, the criminal justice system, and financial markets, their opacity, which hampers humans ability to interpret most of them, has become a problem to be solved. In this work, we present a new taxonomy to classify any method, approach or strategy to deal with the problem of interpretability of machine learning models. The proposed taxonomy fills a gap in the current taxonomy frameworks regarding the subjective perception of different interpreters about the same model. To evaluate the proposed taxonomy, we have classified the contributions of some relevant scientific articles in the area.

With a lineage originating in the days of silent films, visual music, in its current incarnation, is a relatively recent phenomenon when compared to an historically broad field of creative expression. Today it is a time-based audio/visual territory explored and mined by a handful of visual and musical artists. However, an extensive examination of the literature indicates that few of these composers have delved into the associable areas of merging virtual holography and holophony toward visual music composition. It is posited here that such an approach is extremely rich with novel expressive potential and simultaneously with numerous novel challenges. The goal of this study is, through praxis, to instantiate and document an initial exploration into the implementation of holochory toward the creation of visual music compositions. Obviously, engaging holochoric visual music as a means of artistic expression requires an interdisciplinary pipeline. Certainly, this is demonstrated in merging music and visual art into a cohesive form, which is the basis of visual music composition. However, in this study is revealed another form of interdisciplinarity. A major challenge resides with the development of the means to efficiently render the high-resolution stereoscopic images intrinsic to the animation of virtual holograms. Though rendering is a challenge consistent with creating digital animations in general, here the challenge is further exacerbated by the extensive use of multiple reflections and refractions to create complexity from relatively simple geometric objects. This reveals that, with the level of computational technology currently available, the implementation of high-performance computing is the optimal approach. Unifying such diverse areas as music, visual art, and computer science toward a common artistic medium necessitates a methodological approach in which the interdependency between each facet is recognized and engaged. Ultimately, a quadrilateral reciprocative feedback loop, involving the composer's sensibilities in addition to each of the other facets of the compositional process, must be realized in order to facilitate a cohesive methodology leading toward viability. This dissertation provides documentation of methodologies and ideologies undertaken in an initial foray into creating holochoric visual music compositions. Interlaced matrices of contextualization are intended to disseminate the processes involved in deference to composers who will inevitably follow in the wake of this research. Accomplishing such a goal is a quintessential aspect of practice-based research, through which new knowledge is gained during the act of creating. Rather than formulating theoretical perspectives, it is through the praxis of composing holochoric visual music that the constantly arising challenges are recognized, analyzed, and subsequently addressed and resolved in order to ensure progression in the compositional process. Though measuring the success of the resultant compositions is indeed a subjective endeavor, as is the case with all art, the means by which they are achieved is not. The development of such pipelines and processes, and their implementation in practice, are the basic building blocks of further exploration, discovery, and artistic expression. This is the impetus for this document and for my constantly evolving and progressing trajectory as a scholar, artist, composer, and computer scientist. / Doctor of Philosophy / In this paper the author explores the idea that, owing to their shared three-dimensional nature, holophons and holograms are well suited as mediums for visual music composition. This union is ripe with creative opportunity and fraught with challenges in the areas of aesthetics and technical implementation. Squarely situated upon the bleeding edge of phenomenological research and creative practice, this novel medium is nonetheless within reach. Here, one methodological pipeline is delineated that employs the convergence of holography, holophony, and super-computing toward the creation of visual music compositions intended for head mounted displays or large scale 3D/360 projection screens and high-density loudspeaker arrays.

Traitement des mégadonnées, surveillance, prédictions comportementales ou aides à la décision, les avantages techniques et commerciaux attribués à l'intelligence artificielle emportent l’engouement et l’adhésion des acteurs économiques privés. Ayant vocation à reproduire les facultés cognitives de l’être humain, l’intelligence artificielle s’immisce ainsi progressivement dans nos activités, nos usages et plus largement, dans nos vies. Pourtant, les défauts de la technologie inquiètent. Utilisée à des fins de reconnaissance faciale, de profilage publicitaire ou encore de recrutement, les biais de l’intelligence artificielle représentent des risques de discriminations pour les personnes qui interagissent avec cette technologie. Or dans un secteur aussi sensible que le recrutement, un tel risque représente un enjeu aussi bien pour les candidats, exposés à une violation de leur droit fondamental à l’égalité, que pour les employeurs qui, eux, s’exposeraient à des sanctions juridiques. En l’absence d’un cadre juridique spécifique à l’intelligence artificielle, la question se pose donc de savoir si notre droit permet l’appréhension de ces formes de discriminations à l’embauche. Le propos de ce mémoire consistera donc à proposer des réponses à cette interrogation en trois temps : l’étude du cadre légal applicable, la gestion du risque de biais discriminatoire et l’enjeu de l’accès à la justice des candidats lésés. / Big data processing, surveillance, behavioral predictions or decision aids, the technical and commercial advantages attributed to artificial intelligence have won the enthusiasm and support of private economic players. Designed to reproduce the cognitive abilities of human beings, artificial intelligence is gradually interfering in our activities, our practices and more widely in our lives. Nevertheless, the flaws of the technology are concerning. Used for facial recognition, advertising profiling or recruitment purposes, the artificial intelligence biases are a risk of discrimination against people who interact with this technology. In a sector as sensitive as recruitment, such a risk constitutes a challenge both for candidates, exposed to a violation of their fundamental right to equality, and for employers who would be exposed to legal sanctions. Without a specific legal framework for artificial intelligence, the question therefore arises as to whether our law allows for the apprehension of these forms of discrimination in hiring. The purpose of this thesis will consist in proposing answers to this question in three stages : the study of the applicable legal framework, the challenges of managing the risk of discriminatory bias, and access to justice for aggrieved candidates.

Facial recognition technology has become a ubiquitous tool in security and personal identification. However, the rise of this technology has been accompanied by concerns over inherent biases, particularly regarding ethnic and gender. This thesis examines the extent of these biases by focusing on the influence of dataset imbalances in facial recognition algorithms. We employ a structured methodological approach that integrates AI-generated images to enhance dataset diversity, with the intent to balance representation across ethnics and genders. Using the ResNet and Vgg model, we conducted a series of controlled experiments that compare the performance impacts of balanced versus imbalanced datasets. Our analysis includes the use of confusion matrices and accuracy, precision, recall and F1-score metrics to critically assess the model’s performance. The results demonstrate how tailored augmentation of training datasets can mitigate bias, leading to more equitable outcomes in facial recognition technology. We present our findings with the aim of contributing to the ongoing dialogue regarding AI fairness and propose a framework for future research in the field.

Die Nutzung von terrestrischen Ökosystemen zur Befriedigung der Grundbedürfnisse der Menschheit hat tiefgreifende Auswirkungen auf das Erdsystem und führte zur Ausprägung von anthropogen dominierten Landsystemen. Diese sind von hoher Komplexität, da sie aus einer Vielzahl von unterschiedlichsten Einflussfaktoren angetriebenen Landnutzungsveränderungen hervorgegangen sind. Aktuelle Forderungen nach einer nachhaltigen zukünftigen Landnutzung erfordern ein fundiertes und integratives Verständnis dieser Komplexität. Das Hauptziel dieser Arbeit ist es, ein besseres Verständnis der raum-zeitlichen Muster und Determinanten des Landsystemwandels, insbesondere der Landnutzungsintensität, in Europa zwischen 1990 und 2010 zu erlangen. Europa ist ein interessantes Studiengebiet, da es jüngst starke Landnutzungsveränderungen erlebte und seine Heterogenität zu einer Vielfalt von Landsystemen und Landsystemveränderungen führte. Das Ziel der Arbeit wurde durch (i) die Kartierung von Intensitätsmustern und deren Veränderungen in Forst- und Agrarsystemen sowie der Ermittlung der dafür einflussreichsten räumlichen Determinanten und (ii) die Kartierung und Charakterisierung archetypischer Muster und Entwicklungsverläufe von Landsystemen untersucht. Die Ergebnisse dieser Arbeit zeigten einen deutlichen Ost-West-Unterschied in Landsystemmustern und -veränderungen in Europa, mit intensiv genutzten und intensivierenden Regionen vor allem in Westeuropa. Dennoch wurde Europa vor allem durch relativ stabile Landsystemmuster gekennzeichnet und (De-)Intensivierungstrends waren nur von untergeordneter Bedeutung. Intensitätsmuster und -veränderungen waren stark an Standortbedingungen gebunden, vor allem an edaphische, klimatische, und länderspezifische Besonderheiten. Diese Arbeit erweitert das Verständnis des Landsystemwandels in Europa und kann zur Entwicklung wissenschaftlicher und politikbezogener Maßnahmen sowie zur Erreichung einer nachhaltigeren Landnutzung in Europa beitragen. / The utilisation of terrestrial ecosystems to satisfy the basic needs of humankind has profound impacts on the Earth System and led to the development of human-dominated land systems. These are substantially complex as they evolved from a multitude of land-change pathways driven by a variety of influential factors. Current calls for a more sustainable future land-use require a sound and integrative understanding of this complexity. The main goal of this thesis is to better understand the spatio-temporal patterns and the determinants of land-system change in Europe between 1990 and 2010, especially with regard to land-use intensity. Europe serves as an interesting study region as it recently experienced a period of marked land-use change, and since its large environmental, political, and socio-economic heterogeneity resulted in a diversity of land systems and land-change pathways. Land-system changes in Europe were examined by (i) mapping patterns and changes in forestry and agricultural intensity and identifying the most influential spatial determinants related to these changes, and (ii) mapping and characterising archetypical patterns and trajectories of land systems considering both land-use extent and intensity indicators. Results revealed a distinct east-west divide in Europe’s land-system patterns and change trajectories, with intensively used and intensifying regions particularly located in Western Europe. However, Europe was mainly characterised by relatively stable land-systems patterns with (de-) intensification trends being only of minor importance. Land-use intensity levels and changes were strongly related to site conditions, especially with regard to soil and climate, as well as to country-specific characteristics. By fostering the understanding of land-system change, this thesis has the potential to contribute to scientific and policy-related actions that address current efforts to guide future land systems in Europe to a more sustainable use.

Zuweilen wird gefördertes Erdgas als eine Brückentechnologie noch eine Weile erhalten bleiben, aber unsere Gasnetzinfrastruktur hat auch in einer Ära post-fossiler Brennstoffe eine Zukunft, um Klima-neutral erzeugtes Methan, Ammoniak oder Wasserstoff zu transportieren. Damit die Dispatcher der Zukunft, in einer sich fortwährend dynamisierenden Marktsituation, mit sich beständig wechselnden Kleinstanbietern, auch weiterhin einen sicheren Gasnetzbetrieb ermöglichen und garantieren können, werden sie auf moderne, schnelle Simulations- sowie performante Optimierungstechnologie angewiesen sein. Der Schlüssel dazu liegt in einem besseren Verständnis zur numerischen Behandlung nicht differenzierbarer Funktionen und diese Arbeit möchte einen Beitrag hierzu leisten. Wir werden stückweise differenzierbare Funktionen in sog. Abs-Normalen Form betrachten. Durch einen Prozess, der Abs-Linearisierung genannt wird, können wir stückweise lineare Approximationsmodelle erster Ordnung, mittels Techniken der algorithmischen Differentiation erzeugen. Jene Modelle können über Matrizen und Vektoren mittels gängiger Software-Bibliotheken der numerischen linearen Algebra auf Computersystemen ausgedrückt, gespeichert und behandelt werden. Über die Generalisierung der Formel von Faà di Bruno können auch Splinefunktionen höherer Ordnung generiert werden, was wiederum zu Annäherungsmodellen mit besserer Güte führt. Darauf aufbauend lassen sich gemischte Taylor-Kollokationsmethoden, darunter die mit Ordnung zwei konvergente generalisierte Trapezmethode, zur Integration von Gasnetzen, in Form von nicht glatten Algebro-Differentialgleichungssystemen, definieren. Numerische Experimente demonstrieren das Potential. Da solche implizite Integratoren auch nicht lineare und in unserem Falle zugleich auch stückweise differenzierbare Gleichungssysteme erzeugen, die es als Unterproblem zu lösen gilt, werden wir uns auch die stückweise differenzierbare, sowie die stückweise lineare Newtonmethode betrachten. / As of yet natural gas will remain as a bridging technology, but our gas grid infrastructure does have a future in a post-fossil fuel era for the transportation of carbon-free produced methane, ammonia or hydrogen. In order for future dispatchers to continue to enable and guarantee safe gas network operations in a continuously changing market situation with constantly switching micro-suppliers, they will be dependent on modern, fast simulation as well as high-performant optimization technology. The key to such a technology resides in a better understanding of the numerical treatment of non-differentiable functions and this work aims to contribute here. We will consider piecewise differentiable functions in so-called abs-normal form. Through a process called abs-linearization, we can generate piecewise linear approximation models of order one, using techniques of algorithmic differentiation. Those models can be expressed, stored and treated numerically as matrices and vectors via common software libraries of numerical linear algebra. Generalizing the Faà di Bruno's formula yields higher order spline functions, which in turn leads to even higher order approximation models. Based on this, mixed Taylor-Collocation methods, including the generalized trapezoidal method converging with an order of two, can be defined for the integration of gas networks represented in terms of non-smooth system of differential algebraic equations. Numerical experiments will demonstrate the potential. Since those implicit integrators do generate non-linear and, in our case, piecewise differentiable systems of equations as sub-problems, it will be necessary to consider the piecewise differentiable, as well as the piecewise linear Newton method in advance.

Achieving high computation efficiency, in terms of Cycles per Instruction (CPI), for high-performance computing kernels is an interesting and challenging research area. Dense Linear Algebra (DLA) computation is a representative high-performance computing ap- plication, which is used, for example, in LU and QR factorizations. Unfortunately, mod- ern off-the-shelf microprocessors fall significantly short of achieving theoretical lower bound in CPI for high performance computing applications. In this thesis, we perform an in-depth analysis of the available parallelisms and propose suitable algorithmic and architectural variation to significantly improve the computation efficiency. There are two standard approaches for improving the computation effficiency, first, to perform application-specific architecture customization and second, to do algorithmic tuning. In the same manner, we first perform a graph-based analysis of selected DLA kernels. From the various forms of parallelism, thus identified, we design a custom processing element for improving the CPI. The processing elements are used as building blocks for a commercially available Coarse-Grained Reconfigurable Architecture (CGRA). By per- forming detailed experiments on a synthesized CGRA implementation, we demonstrate that our proposed algorithmic and architectural variations are able to achieve lower CPI compared to off-the-shelf microprocessors. We also benchmark against state-of-the-art custom implementations to report higher energy-performance-area product. DLA computations are encountered in many engineering and scientific computing ap- plications ranging from Computational Fluid Dynamics (CFD) to Eigenvalue problem. Traditionally, these applications are written in highly tuned High Performance Comput- ing (HPC) software packages like Linear Algebra Package (LAPACK), and/or Scalable Linear Algebra Package (ScaLAPACK). The basic building block for these packages is Ba- sic Linear Algebra Subprograms (BLAS). Algorithms pertaining LAPACK/ScaLAPACK are written in-terms of BLAS to achieve high throughput. Despite extensive intellectual efforts in development and tuning of these packages, there still exists a scope for fur- ther tuning in this packages. In this thesis, we revisit most prominent and widely used compute bound algorithms like GMM for further exploitation of Instruction Level Parallelism (ILP). We further look into LU and QR factorizations for generalizations and exhibit higher ILP in these algorithms. We first accelerate sequential performance of the algorithms in BLAS and LAPACK and then focus on the parallel realization of these algorithms. Major contributions in the algorithmic tuning in this thesis are as follows: Algorithms: We present graph based analysis of General Matrix Multiplication (GMM) and discuss different types of parallelisms available in GMM We present analysis of Givens Rotation based QR factorization where we improve GR and derive Column-wise GR (CGR) that can annihilate multiple elements of a column of a matrix simultaneously. We show that the multiplications in CGR are lower than GR We generalize CGR further and derive Generalized GR (GGR) that can annihilate multiple elements of the columns of a matrix simultaneously. We show that the parallelism exhibited by GGR is much higher than GR and Householder Transform (HT) We extend generalizations to Square root Free GR (also knows as Fast Givens Rotation) and Square root and Division Free GR (SDFG) and derive Column-wise Fast Givens, and Column-wise SDFG . We also extend generalization for complex matrices and derive Complex Column-wise Givens Rotation Coarse-grained Recon gurable Architectures (CGRAs) have gained popularity in the last decade due to their power and area efficiency. Furthermore, CGRAs like REDEFINE also exhibit support for domain customizations. REDEFINE is an array of Tiles where each Tile consists of a Compute Element and a Router. The Routers are responsible for on-chip communication, while Compute Elements in the REDEFINE can be domain customized to accelerate the applications pertaining to the domain of interest. In this thesis, we consider REDEFINE base architecture as a starting point and we design Processing Element (PE) that can execute algorithms in BLAS and LAPACK efficiently. We perform several architectural enhancements in the PE to approach lower bound of the CPI. For parallel realization of BLAS and LAPACK, we attach this PE to the Router of REDEFINE. We achieve better area and power performance compared to the yesteryear customized architecture for DLA. Major contributions in architecture in this thesis are as follows: Architecture: We present design of a PE for acceleration of GMM which is a Level-3 BLAS operation We methodically enhance the PE with different features for improvement in the performance of GMM For efficient realization of Linear Algebra Package (LAPACK), we use PE that can efficiently execute GMM and show better performance For further acceleration of LU and QR factorizations in LAPACK, we identify macro operations encountered in LU and QR factorizations, and realize them on a reconfigurable data-path resulting in 25-30% lower run-time

Cette thèse développe l’analyse d’une politique économique environnementale appliquée dans le cadre d’une économie ouverte et dans celui d’une économie fermée. Elle étudie les effets sur le bien-être et l’environnement des réglementations volontaire et obligatoire tenant compte de l’hétérogénéité des agents économiques et des pays. Elle s’intéresse en particulier aux différents types d’éco-étiquetages en autarcie (Chapitre 1) et après ouverture au commerce international (Chapitre 2); aux programmes d’apport d’information à plusieurs niveaux (Chapitre 3); et aux taxes sur les émissions et marchés publics écologiques (Chapitre 4). L’analyse montre que non seulement le gouvernement mais également les éco-consommateurs peuvent inciter même les producteurs éco-indifférents à se décider pour l’agir respectueux de l’environnement. Les instruments de politique environnementale induisent auto-sélection et polarisation dans les marchés desservis par des entreprises hétérogènes en termes de productivité. Je démontre que des instruments volontaires peuvent avoir des résultats positifs sur le bien-être et l’environnement. En fonction d’hypothèses précises, ils peuvent également être plus efficaces que des approches obligatoires. Le modèle montre également que quand il y a ouverture au commerce international, la politique écologique a un effet supplémentaire sur le bien-être et sur l’environnement, dépendamment du type de politique et des consciences environnementales des différents pays commerciaux. / This thesis develops an applied environmental economic policy analysis in closed and open economy frameworks. It investigates welfare and environmental outcomes of voluntary and mandatory regulation allowing for heterogeneity across economic agents and countries. Particularly, it focuses on voluntary eco-labels of different types in autarky (Chapter 1) and upon opening to international trade (Chapter 2); multi-tier information provision programmes (Chapter 3); and emission taxes and green public procurement (Chapter 4). The analysis shows that not only the government but also eco-concerned consumers can incentivise even eco-indifferent producers to act more environmentally-friendly. Environmental policy instruments induce self-selection and polarisation in the markets served by firms heterogeneous in their productivity. I demonstrate that voluntary instruments can lead to positive welfare and environmental outcomes. Under particular assumptions, they also can be more efficient than mandatory approaches. The model also shows that upon opening to international trade eco-policy yields additional welfare and environmental effects conditionally on the type of the policy and the environmental awareness difference across trading countries.

Predicting the behavior of financial markets is largely an unsolved problem. The problem hasbeen approached with many different methods ranging from binary logic, statisticalcalculations and genetic algorithms. In this thesis, the problem is approached with a machinelearning method, namely the Long Short-Term Memory (LSTM) variant of Recurrent NeuralNetworks (RNNs). Recurrent neural networks are artificial neural networks (ANNs)—amachine learning algorithm mimicking the neural processing of the mammalian nervoussystem—specifically designed for time series sequences. The thesis investigates the capabilityof the LSTM in modeling financial market behavior as well as compare it to the traditionalRNN, evaluating their performances using various measures. / Prediktion av den finansiella marknadens beteende är i stort ett olöst problem. Problemet hartagits an på flera sätt med olika metoder så som binär logik, statistiska uträkningar ochgenetiska algoritmer. I den här uppsatsen kommer problemet undersökas medmaskininlärning, mer specifikt Long Short-Term Memory (LSTM), en variant av rekurrentaneurala nätverk (RNN). Rekurrenta neurala nätverk är en typ av artificiellt neuralt nätverk(ANN), en maskininlärningsalgoritm som ska efterlikna de neurala processerna hos däggdjursnervsystem, specifikt utformat för tidsserier. I uppsatsen undersöks kapaciteten hos ett LSTMatt modellera finansmarknadens beteenden och jämförs den mot ett traditionellt RNN, merspecifikt mäts deras effektivitet på olika vis.

Les files d'attente fluides sont des processus markoviens à deux dimensions, où la première composante, appelée le niveau, représente le contenu d'un réservoir et prend des valeurs continues, et la deuxième composante, appelée la phase, est l'état d'un processus markovien dont l'évolution contrôle celle du niveau. Le niveau de la file fluide varie linéairement avec un taux qui dépend de la phase et qui peut prendre n'importe quelle valeur réelle.<p><p>Dans cette thèse, nous explorons le lien entre les files fluides et les processus QBD, et nous appliquons des arguments utilisés en théorie des processus de renouvellement pour obtenir la distribution stationnaire de plusieurs modèles fluides.<p><p>Nous commençons par l'étude d'une file fluide avec un réservoir de taille infinie; nous déterminons sa distribution stationnaire, et nous présentons un algorithme permettant de calculer cette distribution de manière très efficace. Nous observons que la distribution stationnaire de la file fluide de capacité infinie est très semblable à celle d'un processus QBD avec une infinité de niveaux. Nous poursuivons la recherche des similarités entre les files fluides et les processus QBD, et nous étudions ensuite la distribution stationnaire d'une file fluide de capacité finie. Nous montrons que l'algorithme valable pour le cas du réservoir infini permet de calculer toutes les quantités importantes du modèle avec un réservoir fini.<p><p>Nous considérons ensuite des modèles fluides plus complexes, de capacité finie ou infinie, où le comportement du processus markovien des phases peut changer lorsque le niveau du réservoir atteint certaines valeurs seuils. Nous montrons que les méthodes développées pour des modèles classiques s'étendent de manière naturelle à ces modèles plus complexes.<p><p>Pour terminer, nous étudions les conditions nécessaires et suffisantes qui mènent à l'indépendance du niveau et de la phase d'une file fluide de capacité infinie en régime stationnaire. Ces résultats s'appuient sur des résultats semblables concernant des processus QBD.<p><p>Markov modulated fluid queues are two-dimensional Markov processes, of which the first component, called the level, represents the content of a buffer or reservoir and takes real values; the second component, called the phase, is the state of a Markov process which controls the evolution of the level in the following manner: the level varies linearly at a rate which depends on the phase and which can take any real value.<p><p>In this thesis, we explore the link between fluid queues and Quasi Birth-and-Death (QBD) processes, and we apply Markov renewal techniques in order to derive the stationary distribution of various fluid models.<p><p>To begin with, we study a fluid queue with an infinite capacity buffer; we determine its stationary distribution and we present an algorithm which performs very efficiently in the determination of this distribution. We observe that the equilibrium distribution of the fluid queue is very similar to that of a QBD process with infinitely many levels. We further exploit the similarity between the two processes, and we determine the stationary distribution of a finite capacity fluid queue. We show that the algorithm available in the infinite case allows for the computation of all the important quantities entering in the expression of this distribution.<p><p>We then consider more complex models, of either finite or infinite capacities, in which the behaviour ff the phase process may change whenever the buffer is empty or full, or when it reaches certain thresholds. We show that the techniques that we develop for the simpler models can be extended quite naturally in this context.<p><p>Finally, we study the necessary and sufficient conditions that lead to the independence between the level and the phase of an infinite capacity fluid queue in the stationary regime. These results are based on similar developments for QBD processes. / Doctorat en sciences, Spécialisation mathématiques / info:eu-repo/semantics/nonPublished