Дисертації з теми "Génératif"
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-50 дисертацій для дослідження на тему "Génératif".
Біля кожної праці в переліку літератури доступна кнопка «Додати до бібліографії». Скористайтеся нею – і ми автоматично оформимо бібліографічне посилання на обрану працю в потрібному вам стилі цитування: APA, MLA, «Гарвард», «Чикаго», «Ванкувер» тощо.
Також ви можете завантажити повний текст наукової публікації у форматі «.pdf» та прочитати онлайн анотацію до роботи, якщо відповідні параметри наявні в метаданих.
Переглядайте дисертації для різних дисциплін та оформлюйте правильно вашу бібліографію.
Loyauté, Gautier. "Un modèle génératif pour le développement de serveurs Internet." Phd thesis, Université Paris-Est, 2008. http://tel.archives-ouvertes.fr/tel-00470539.
Повний текст джерелаCao, Yi-Heng. "Apprentissage génératif pour la synthèse d'images médicales dynamiques 4D." Electronic Thesis or Diss., Brest, 2024. http://www.theses.fr/2024BRES0004.
Повний текст джерелаFour-dimensional computed tomography (4DCT) involves reconstructing an acquisition in multiple phases to track the movements of internal organs and tumors. It is used routinely for radiotherapy treatment planning of lung cancer, but it exposes patients to higher radiation doses, up to six times greater than those of a conventional threedimensional computed tomography (3DCT). Deep learning methods from the field of computer vision are gaining significant interest within the medical imaging community. Among these approaches, generative models stand out due to their ability to generate synthetic images that faithfully replicate the appearance and statistical characteristics of images acquired from real systems. In this thesis, we explore the use of a generative model for dynamic image generation. We propose a model capable of generating patient-specific respiratory motion from a diagnostic 3DCT image and respiratory data. The goal is to enable radiologists to delineate target volumes and organs at risk, as well as perform dose calculations on these dynamic synthetic images. This method would reduce the need for a 4DCT acquisition, thereby reducing the patient’s radiation exposure
Krauth, Timothé. "Modèle génératif profond pour l'estimation de probabilité de collision en vol." Electronic Thesis or Diss., Toulouse, ISAE, 2024. http://www.theses.fr/2024ESAE0018.
Повний текст джерелаIt is essential to calculate the probability of aircraft collisions to optimise air traffic while maintaining high safety standards. This need became more pronounced in the 1960s with the increase in transatlantic commercial air traffic. Initially, analytical models such as those of Reich and Anderson-Hsu were benchmarks for assessing in-flight collision risks, but they proved to be less suited for the complex airspace around airports.Data-driven methods, especially Monte Carlo simulations, have become a promising alternative for collision risk assessment. They offer significant flexibility through simplified assumptions, making them adaptable to various contexts. However, traditional Monte Carlo simulations are inefficient for estimating rare event probabilities, requiring a large number of aircraft trajectories and substantial computational resources. This thesis proposes a collision risk model based on Monte Carlo simulations, using a trajectory generation model to overcome these limitations associated with rare events. These generative methods faithfully reproduce observed trajectory distributions while incorporating uncertainties from external factors. Three main research areas are defined: (i) developing a trajectory generation method, (ii) constructing a Monte Carlo-based collision risk model using synthetic trajectories, and (iii) improving the interpretability of collision risk estimates.Generating synthetic samples involves estimating the distribution of observed data to ensure identical distribution in new samples. This is particularly important for aircraft trajectories, where the model must reflect uncertainty sources causing deviations from standard trajectories. We initially use traditional statistical learning methods to estimate complex two-dimensional aircraft trajectories. Despite reducing the problem's dimensionality, conventional methods struggle with high-dimensional distribution estimation. We then explore the use of variational autoencoders for more refined probability density estimation. Suitably adapted for multivariate time-series applications, variational autoencoders prove effective for estimating the distribution of complex aircraft trajectories.Using the developed generation method, we estimate the risk of loss of separation induced by the departure and approach procedures of Paris-Orly Airport using Monte Carlo simulations. The use of a trajectory generation method proves promising, allowing the creation of the equivalent of 20 years of air traffic trajectories from only two months of observations. However, this direct method has limitations for estimating extremely low collision probabilities, requiring the use of one variational autoencoder per flight procedure considered in the studied scenario. The processes of trajectory generation and collision risk evaluation are distinctly separated. Consequently, the inherent constraints of classical Monte Carlo methods are not truly overcome but merely postponed by the production of a set of arbitrarily large trajectories.The thesis's final work unifies the frameworks of variational autoencoders and uncertainty quantification. It demonstrates how variational autoencoders can build suitable input distributions for uncertainty quantification algorithms, enhancing the reliability of Monte Carlo simulations through subset simulation and the explainability of mid-air collision probability estimation through sensitivity analysis. More broadly, we show that the variational autoencoder represents a promising tool to be associated with uncertainty quantification problems
Jean, Fabien. "L'échange génératif de technologies innovantes : engagement conceptif et conception de la valeur." Thesis, Paris Sciences et Lettres (ComUE), 2016. http://www.theses.fr/2016PSLEM082/document.
Повний текст джерелаInnovation processes of large companies experience a lack of resources between Research and Development, i.e. the valley of death. Safran created its Innovation Department to counterbalance. It is based on exchanges with the subsidiaries. However such exchanges cannot be processed in most early stages, when they include unknowns. For instance, ideas generated through the DKCP method remain unexplored. This intervention-research aims at establishing a model of collective action to exchange in the unknown. It defends the thesis that, in the unknown, sellers and buyers of innovative technologies engage in the design of the boundary between technologies and environments.Facing the limits of classic models of exchanges of economy, decision, intéressement and creativity, this thesis adopts a design-theories framework. It proposes the model of design engagement of resources. It explains the limits of a common tool for exchanging technologies, i.e. Technology Readiness Levels. It proposes to model exchanges between seller and buyer as the exploration of the Technology-Environment boundary. To do so it constructs the C-K T / C-K E formalism. Finally, the method Steering Exploration Through Technology and Environment Invariants ("Pilotage de l'Exploration Par les Invariants Technologie-Environnement" (PEPITE) in French) is constructed in collaboration with Safran Innovation Department. It is based on two cases of explorations which passed the valley of death within the researchers intervention
Sallé, Guillaume. "Apprentissage génératif à bas régime des données pour la segmentation d'images en oncologie." Electronic Thesis or Diss., Brest, 2024. http://www.theses.fr/2024BRES0032.
Повний текст джерелаIn statistical learning, the performance of models is affected by various biases present within the data, including data scarcity and domain shift. This thesis focuses on reducing their impact in the field of pathological structure segmentation in medical imaging. Our goal is to minimize data discrepancies at the region of interest (ROI) level between the training source domain and the target deployment domain, whether it is intrinsic to the data or caused by the limited data availability. To this end, we present an adaptive data augmentation strategy, based on the analysis of the intensity distribution of the ROIs in the deployment domain.A first contribution, which we call naive augmentation, consists of altering the appearance of the training ROIs to better match the characteristics of the ROIs in the deployment domain. A second augmentation, complementing the first, makes the alteration more realistic relative to the properties of the target domain by harmonizing the characteristics of the altered image. For this, we employ a generative model trained on a single unlabeled image from the deployment domain (one-shot approach), making this technique usable in any data regime encountered. In this way, we enhance the robustness of the downstream segmentation model forROIs whose characteristics were initially underrepresented in the deployment domain. The effectiveness of this method is evaluated under various data regimes and in different clinical contexts (MRI, CT, CXR). Our approach demonstrated impressive results in a tumor segmentation challenge at MICCAI 2022
Beller, Grégory. "Analyse et modèle génératif de l'expressivité : application à la Parole et à l'Interprétation musicale." Phd thesis, Université Pierre et Marie Curie - Paris VI, 2009. http://tel.archives-ouvertes.fr/tel-00431104.
Повний текст джерелаClair, Romain. "Etude de méthodes de production d'art génératif et de leur application pour la conception d'outils de création artistique accessibles." Thesis, Tours, 2010. http://www.theses.fr/2010TOUR4026/document.
Повний текст джерелаStarting from a survey of algorithmic methods for automatic artistic creation, our works deals with adaptations of these methods to provide accessible art creation forms, focusing on music and computer drawing.the development of accessible computer software requires some constraints. Following these constraints and using adapted automatic art production systems, notably based on artificial ant colony algorithms, we developed two computer programs.The first one is a virtual music instrument, allowing most people to play music and providing and automatic accompaniment.The second one is a drawing workshop with generative methods-based tools provide complex results from simple actions.This PhD thesis details the development of this two programs and their evaluations, with real users meetings
Schwartzmann, Matheus Nogueira. "Cartas marcadas : prática epistolar e formas de vida na correspondência de Mário de Sá-Carneiro /." Araraquara : [s.n.], 2009. http://hdl.handle.net/11449/103560.
Повний текст джерелаBanca: Maria de Lourdes Ortiz Gandini Baldan
Banca: Loredana Limoli
Banca: Maria Celia de Moraes Leonel
Banca: Vera Lucia Rodella Abriata
Resumo: A partir da análise das cartas enviadas pelo poeta português Mário de Sá-Carneiro ao poeta Fernando Pessoa, busca-se traçar, primeiramente, a arquitetura da correspondência, para somente então identificar a própria constituição do "sujeito Mário de Sá-Carneiro" nela instaurado. Para tanto, emprega-se o instrumental teórico-metodológico da semiótica de inspiração greimasiana, especialmente a noção de "forma de vida" e as proposições mais recentes de Jacques Fontanille, como o estudo das "práticas semióticas" e a sua proposta de um "percurso gerativo da expressão". Toma-se, assim, de início, a carta-objeto em sua materialidade, explicitando-se as coerções de ordem espacial e material que sofre enquanto objeto-suporte, as marcas que outras práticas (o sistema postal, por exemplo) nela inserem e a sua organização topológica. Na sequência, observa-se como as propriedades textuais e discursivas organizam-se no interior da carta e como são organizadas as próprias cartas no interior da correspondência, por meio de estratégias contínuas (como a sinceridade fiduciária) que lhe dão ritmo e garantem a sua existência enquanto prática epistolar eficiente. Do ponto de vista da identidade dos sujeitos inscritos nas cartas, busca-se, a princípio, identificar os espaços que habitam, sendo Paris o que mais intensamente é manifestado, e qual é o tempo (cronológico e afetivo) em que vivem a sua amizade epistolar. Em um segundo momento, apresentam-se as formas afetivas produzidas pelos percursos passionais que o sujeito Sá-Carneiro assume e pelas estratégias de veridicção que a carta instaura (como a sinceridade e a verdade, a espera e o desespero, a saudade e a angústia), que, reiteradas continuamente, representam verdadeiras "formas de vida" que definem a identidade do sujeito epistolar sá-carneiriano.
Résumé: A partir de l'analyse des lettres envoyées par le poète portugais Mário de Sá-Carneiro au poète Fernando Pessoa, on cherche à dégager, d'abord, l'architecture de la correspondance pour identifier, ensuite, la constitution même du « sujet Mário de Sá-Carneiro » qui y est instauré. Pour ce faire, on emploie ici la sémiotique d'inspiration greimassienne, plus précisément la notion de « formes de vie » et les contributions théoriques les plus récentes de Jacques Fontanille, notamment ces réflexions sur les « pratiques sémiotiques » et sur le « parcours génératif de l'expression ». Tout d'abord, on s'occupe de la « lettre-objet » dans toute sa matérialité, en mettant en évidence les contraintes spatiales et matérielles qu'elle subit en tant qu'objet-support ; les empreintes que toutes les autres pratiques lui impriment (le système postal, notamment) ; et aussi sa propre structure topologique. Ensuite, on examine l'organisation des propriétés textuelles et discursives au sein de la lettre et l'organisation même des lettres dans le contexte de la correspondance, à travers des stratégies persistantes (tels que la « sincérité fiduciaire ») qui, en donnant du rythme à la correspondance, assurent son existence en tant que pratique épistolaire efficiente. Du point de vie de l'identité des sujets inscrits dans les lettres analysées, on cherche, d'abord, à identifier les espaces qui les deux interlocuteurs habitent - Paris en étant l'espace le plus intensément manifesté - et le temps (chronologique et aussi affectif) où se déroule son amitié épistolaire. Ensuite, on met en valeur toutes les formes affectives (la sincérité et la vérité ; l'attente et le désespoir ; la nostalgie et l'angoisse) qui surgissent à partir des stratégies véridictoires que la lettre convoque et des parcours passionnels assumés pleinement par le sujet Sá-Carneiro.
Doutor
Gaillard, Pierre. "Apprentissage statistique de la connexité d'un nuage de points par modèle génératif : application à l'analyse exploratoire et la classification semi-supervisée." Compiègne, 2008. http://www.theses.fr/2008COMP1767.
Повний текст джерелаIn this work, we propose a statistical model to learn the connectedness of a set of points. This model combine geometrical and statistical approaches by defining a mixture model based on a graph. From this generative graph, we propose and evaluate methods and algorithms to analyse the set of points and to realize semi-supervised learning
Azeraf, Elie. "Classification avec des modèles probabilistes génératifs et des réseaux de neurones. Applications au traitement des langues naturelles." Electronic Thesis or Diss., Institut polytechnique de Paris, 2022. https://theses.hal.science/tel-03880848.
Повний текст джерелаMany probabilistic models have been neglected for classification tasks with supervised learning for several years, as the Naive Bayes or the Hidden Markov Chain. These models, called generative, are criticized because the induced classifier must learn the observations' law. This problem is too complex when the number of observations' features is too large. It is especially the case with Natural Language Processing tasks, as the recent embedding algorithms convert words in large numerical vectors to achieve better scores.This thesis shows that every generative model can define its induced classifier without using the observations' law. This proposition questions the usual categorization of the probabilistic models and classifiers and allows many new applications. Therefore, Hidden Markov Chain can be efficiently applied to Chunking and Naive Bayes to sentiment analysis.We go further, as this proposition allows to define the classifier induced from a generative model with neural network functions. We "neuralize" the models mentioned above and many of their extensions. Models so obtained allow to achieve relevant scores for many Natural Language Processing tasks while being interpretable, able to require little training data, and easy to serve
Hadjeres, Gaëtan. "Modèles génératifs profonds pour la génération interactive de musique symbolique." Thesis, Sorbonne université, 2018. http://www.theses.fr/2018SORUS027/document.
Повний текст джерелаThis thesis discusses the use of deep generative models for symbolic music generation. We will be focused on devising interactive generative models which are able to create new creative processes through a fruitful dialogue between a human composer and a computer. Recent advances in artificial intelligence led to the development of powerful generative models able to generate musical content without the need of human intervention. I believe that this practice cannot be thriving in the future since the human experience and human appreciation are at the crux of the artistic production. However, the need of both flexible and expressive tools which could enhance content creators' creativity is patent; the development and the potential of such novel A.I.-augmented computer music tools are promising. In this manuscript, I propose novel architectures that are able to put artists back in the loop. The proposed models share the common characteristic that they are devised so that a user can control the generated musical contents in a creative way. In order to create a user-friendly interaction with these interactive deep generative models, user interfaces were developed. I believe that new compositional paradigms will emerge from the possibilities offered by these enhanced controls. This thesis ends on the presentation of genuine musical projects like concerts featuring these new creative tools
Hadjeres, Gaëtan. "Modèles génératifs profonds pour la génération interactive de musique symbolique." Electronic Thesis or Diss., Sorbonne université, 2018. http://www.theses.fr/2018SORUS027.
Повний текст джерелаThis thesis discusses the use of deep generative models for symbolic music generation. We will be focused on devising interactive generative models which are able to create new creative processes through a fruitful dialogue between a human composer and a computer. Recent advances in artificial intelligence led to the development of powerful generative models able to generate musical content without the need of human intervention. I believe that this practice cannot be thriving in the future since the human experience and human appreciation are at the crux of the artistic production. However, the need of both flexible and expressive tools which could enhance content creators' creativity is patent; the development and the potential of such novel A.I.-augmented computer music tools are promising. In this manuscript, I propose novel architectures that are able to put artists back in the loop. The proposed models share the common characteristic that they are devised so that a user can control the generated musical contents in a creative way. In order to create a user-friendly interaction with these interactive deep generative models, user interfaces were developed. I believe that new compositional paradigms will emerge from the possibilities offered by these enhanced controls. This thesis ends on the presentation of genuine musical projects like concerts featuring these new creative tools
Schwartzmann, Matheus Nogueira [UNESP]. "Cartas marcadas: prática epistolar e formas de vida na correspondência de Mário de Sá-Carneiro." Universidade Estadual Paulista (UNESP), 2009. http://hdl.handle.net/11449/103560.
Повний текст джерелаCoordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)
A partir de l’analyse des lettres envoyées par le poète portugais Mário de Sá-Carneiro au poète Fernando Pessoa, on cherche à dégager, d’abord, l’architecture de la correspondance pour identifier, ensuite, la constitution même du « sujet Mário de Sá-Carneiro » qui y est instauré. Pour ce faire, on emploie ici la sémiotique d’inspiration greimassienne, plus précisément la notion de « formes de vie » et les contributions théoriques les plus récentes de Jacques Fontanille, notamment ces réflexions sur les « pratiques sémiotiques » et sur le « parcours génératif de l’expression ». Tout d’abord, on s’occupe de la « lettre-objet » dans toute sa matérialité, en mettant en évidence les contraintes spatiales et matérielles qu’elle subit en tant qu’objet-support ; les empreintes que toutes les autres pratiques lui impriment (le système postal, notamment) ; et aussi sa propre structure topologique. Ensuite, on examine l’organisation des propriétés textuelles et discursives au sein de la lettre et l’organisation même des lettres dans le contexte de la correspondance, à travers des stratégies persistantes (tels que la « sincérité fiduciaire ») qui, en donnant du rythme à la correspondance, assurent son existence en tant que pratique épistolaire efficiente. Du point de vie de l’identité des sujets inscrits dans les lettres analysées, on cherche, d’abord, à identifier les espaces qui les deux interlocuteurs habitent — Paris en étant l’espace le plus intensément manifesté — et le temps (chronologique et aussi affectif) où se déroule son amitié épistolaire. Ensuite, on met en valeur toutes les formes affectives (la sincérité et la vérité ; l’attente et le désespoir ; la nostalgie et l’angoisse) qui surgissent à partir des stratégies véridictoires que la lettre convoque et des parcours passionnels assumés pleinement par le sujet Sá-Carneiro.
A partir da análise das cartas enviadas pelo poeta português Mário de Sá-Carneiro ao poeta Fernando Pessoa, busca-se traçar, primeiramente, a arquitetura da correspondência, para somente então identificar a própria constituição do “sujeito Mário de Sá-Carneiro” nela instaurado. Para tanto, emprega-se o instrumental teórico-metodológico da semiótica de inspiração greimasiana, especialmente a noção de “forma de vida” e as proposições mais recentes de Jacques Fontanille, como o estudo das “práticas semióticas” e a sua proposta de um “percurso gerativo da expressão”. Toma-se, assim, de início, a carta-objeto em sua materialidade, explicitando-se as coerções de ordem espacial e material que sofre enquanto objeto-suporte, as marcas que outras práticas (o sistema postal, por exemplo) nela inserem e a sua organização topológica. Na sequência, observa-se como as propriedades textuais e discursivas organizam-se no interior da carta e como são organizadas as próprias cartas no interior da correspondência, por meio de estratégias contínuas (como a sinceridade fiduciária) que lhe dão ritmo e garantem a sua existência enquanto prática epistolar eficiente. Do ponto de vista da identidade dos sujeitos inscritos nas cartas, busca-se, a princípio, identificar os espaços que habitam, sendo Paris o que mais intensamente é manifestado, e qual é o tempo (cronológico e afetivo) em que vivem a sua amizade epistolar. Em um segundo momento, apresentam-se as formas afetivas produzidas pelos percursos passionais que o sujeito Sá-Carneiro assume e pelas estratégias de veridicção que a carta instaura (como a sinceridade e a verdade, a espera e o desespero, a saudade e a angústia), que, reiteradas continuamente, representam verdadeiras “formas de vida” que definem a identidade do sujeito epistolar sá-carneiriano.
Mery, Bruno. "Modélisation de la Sémantique Lexicale dans le cadre de la théorie des types." Phd thesis, Université Sciences et Technologies - Bordeaux I, 2011. http://tel.archives-ouvertes.fr/tel-00627432.
Повний текст джерелаToumi, Mohsen. "Processus créatifs et systèmes auto-génératifs : automatisation, auto-génération et énaction : esquisse d'une esthétique énactive." Thesis, Toulouse 2, 2016. http://www.theses.fr/2016TOU20090.
Повний текст джерелаThe purpose of this study is the creative processes of automatic and self-generating systems. The raised issue is based on the following paradox: how is it possible for an automatic and autopoietic system to create something that has an aesthetic value? Can we speak of poietic and aesthetic that are characteristics of self-generating systems? If possible, in what sense it is. And what would the implications and the stakes be. How to reconcile consciousness, intentionality emotion, with automation and algorithmic programming? Based on a sample of nine artists, and from a deductive-inductive approach, we have tried to reflect on the necessary links between: automation, autonomy, self-generation and énaction. We have come to realize that there is a continuous evolution that runs the entire history of the instrumentalization of art. This mediation, always evolving, has eventually led to a qualitative change. That is to a passage from a paradigm of do to a paradigm of make do with which the processes of automation, self-generation and enaction have become increasing crucial to the act of artistic creation. The make do paradigm is a paradigm of autopoiesis and emergence, in which, the artist no longer generates directly the work of art, but rather establishes the context of its self-generation. This progressive self-generation takes multiple forms and degrees, beginning with the combinatorial, explaratory, adaptive and transformational generativity through which the traditional creative triangle is found inoperative, and therefore the theory of advance art has changed its categories. In this context of paradigmatic change, à “new hybridization Human-Machine”, a “collective and distributed creativity” or “ecosystem auto-generativity” have become possible leads to new enactive and creative practices, to new adventures of aesthetic theory
Roca, Vincent. "Harmonisation multicentrique d'images IRM du cerveau avec des modèles génératifs non-supervisés." Electronic Thesis or Diss., Université de Lille (2022-....), 2023. http://www.theses.fr/2023ULILS060.
Повний текст джерелаMagnetic resonance imaging (MRI) enables the acquisition of brain images used in the study of neurologic and psychiatric diseases. MR images are more and more used in statistical studies to identify biomarkers and for predictive models. To improve statistical power, these studies sometimes pool data acquired with different machines, which may introduce technical variability and bias into the analysis of biological variabilities. In the last few years, harmonization methods have been proposed to limit the impact of these variabilities. Many studies have notably worked on generative models based on unsupervised deep learning. The doctoral research is within the context of these models, which constitute a promising but still exploratory research field. In the first part of this manuscript, a review of the prospective harmonization methods is proposed. Different methods consisting in normalization applied at the image level, domain translation or style transfer are described to understand their respective issues, with a special focus on unsupervised generative models. The second part is about the methods for evaluation of retrospective harmonization. A review of these methods is first conducted. The most common rely on “traveling” subjects to assume ground truths for harmonization. The review also presents evaluations employed in the absence of such subjects: study of inter-domain differences, biological patterns and performances of predictive models. Experiments showing limits of some approaches commonly employed and important points to consider for their use are then proposed. The third part presents a new model for harmonization of brain MR images based on a CycleGAN architecture. In contrast with the previous works, the model is three-dimensional and processes full volumes. MR images from six datasets that vary in terms of acquisition parameters and age distributions are used to test the method. Analyses of intensity distributions, brain volumes, image quality metrics and radiomic features show an efficient homogenisation between the different sites of the study. Next, the conservation and the reinforcement of biological patterns are demonstrated with an analysis of the evolution of gray-matter volume estimations with age, experiments of age prediction, ratings of radiologic patterns in the images and a supervised evaluation with a traveling subject dataset. The fourth part also presents an original harmonization method with major updates of the first one in order to establish a “universal” generator able to harmonize images without knowing their domain of origin. After a training with data acquired on eleven MRI scanners, experiments on images from sites not seen during the training show a reinforcement of brain patterns relative to age and Alzheimer after harmonization. Moreover, comparisons with other intensity harmonization approaches suggest that the model is more efficient and more robust to different tasks subsequent to harmonization. These different works are a significant contribution to the domain of retrospective harmonization of brain MR images. The bibliographic documentations indeed provide a methodological knowledge base for the future studies in this domain, whether for harmonization in itself or for validation. In addition, the two developed models are two robust tools publicly available that may be integrated in future MRI multicenter studies
Sébillot, Pascale. "Apprentissage sur corpus de relations lexicales sémantiques - La linguistique et l'apprentissage au service d'applications du traitement automatique des langues." Habilitation à diriger des recherches, Université Rennes 1, 2002. http://tel.archives-ouvertes.fr/tel-00533657.
Повний текст джерелаDo, Phan-Thuan. "Arbres de génération et génération exhaustive." Dijon, 2008. http://www.theses.fr/2008DIJOS027.
Повний текст джерелаThe work presented in my thesis is one of research results recently published by the Combinatorial Algorithmic group belonging to the laboratory LE2I, University of Burgundy, France. The aim of the thesis is to explore systematically the technique of generating trees in the context of the exhaustive generation of combinatorial objects. More precisely, it relies on the method of Enumerating Combinatorial Objects (ECO) first proposed by Barcucci et al. 1999. We study about the exhaustive generation of combinatorial objects based on generating trees for these classes, in order to construct efficient algorithms, in a representation and a natural order. First, we introduce a study on the generation of Dyck words and several relative classes of words represented in the lattice path Z2. We give efficient exhaustive generating algorithms for each class under consideration. All of the algorithms presented here are based on the unified approach of the ECO-method. Then we present a new technique, succession function, which refine succession rule. It permits to easily construct general efficient generating algorithms for these classes and to find some new classes easier than previous methods. The last parts study deeply on well-known classes which are generalised Fibonacci and Lucas classes and classes of compositions. New succession rules are proposed. New classes of pattern avoiding permutations are established and put in bijection with those classes
El, Jed Olfa. "WebSum : système de résumé automatique de réponses des moteurs de recherche." Toulouse 3, 2006. http://www.theses.fr/2006TOU30145.
Повний текст джерелаThis thesis lies within the general framework of the information retrieval and more precisely, within the framework of the web document classification and organization. Our objective is to develop a system of automatic summarizing of the search engine answers in the encyclopaedic style (WebSum). This type of summary aims at classifying the search engine answers according to the various topics or what we call in our work, facets of the user query. To carry out this objective, we propose : - A method of identification of the facets of a given query based on the generative lexicon; - An approach of classification of the search engine answers under this various facets; - And a method of evaluation of the relevance of the web pages
Gallas, Mohamed-Anis. "De l'intention à la solution architecturale : proposition d'une méthode d'assistance à la prise en compte de la lumière naturelle durant les phases amont de conception." Thesis, Université de Lorraine, 2013. http://www.theses.fr/2013LORR0101/document.
Повний текст джерелаThe daylight illuminates the architectural space and creates the necessary conditions to accommodate human activities. The illuminating function of daylight is joined to another sensitive dimension that attributes a distinctive identity to the designed space. The control of daylight in architectural environment needs some detailed and precise characteristics about aperture and walls features. The multiplicity of these characteristics is faced with the inaccuracy and the uncertainty of the design information available during the early steps. Our research aims to propose a design support method that takes into account the early design step features and helps designers to integrate there daylight atmosphere intentions in project. We propose a design support method structured as a declarative modelling process. The declarative process helps the designer to declare his daylight intentions and to translate them to potential solutions that could be integrated in his project. This method considers the designer intentions as the main design information used to help designer during the early design steps. The proposed method provides functionalities that could operate the inaccuracy and the uncertainty of the ideas research and formalisation steps. Our design support method was implemented in a prototype design support tool. The cognitive contribution and the ability to support the design activities of the proposed tool was evaluated and analysed in an experimental design context
Prang, Mathieu. "Representation learning for symbolic music." Electronic Thesis or Diss., Sorbonne université, 2021. http://www.theses.fr/2021SORUS489.
Повний текст джерелаA key part in the recent success of deep language processing models lies in the ability to learn efficient word embeddings. These methods provide structured spaces of reduced dimensionality with interesting metric relationship properties. These, in turn, can be used as efficient input representations for handling more complex tasks. In this thesis, we focus on the task of learning embedding spaces for polyphonic music in the symbolic domain. To do so, we explore two different approaches.We introduce an embedding model based on a convolutional network with a novel type of self-modulated hierarchical attention, which is computed at each layer to obtain a hierarchical vision of musical information.Then, we propose another system based on VAEs, a type of auto-encoder that constrains the data distribution of the latent space to be close to a prior distribution. As polyphonic music information is very complex, the design of input representation is a crucial process. Hence, we introduce a novel representation of symbolic music data, which transforms a polyphonic score into a continuous signal.Finally, we show the potential of the resulting embedding spaces through the development of several creative applications used to enhance musical knowledge and expression, through tasks such as melodies modification or composer identification
Sadok, Samir. "Audiovisual speech representation learning applied to emotion recognition." Electronic Thesis or Diss., CentraleSupélec, 2024. http://www.theses.fr/2024CSUP0003.
Повний текст джерелаEmotions are vital in our daily lives, becoming a primary focus of ongoing research. Automatic emotion recognition has gained considerable attention owing to its wide-ranging applications across sectors such as healthcare, education, entertainment, and marketing. This advancement in emotion recognition is pivotal for fostering the development of human-centric artificial intelligence. Supervised emotion recognition systems have significantly improved over traditional machine learning approaches. However, this progress encounters limitations due to the complexity and ambiguous nature of emotions. Acquiring extensive emotionally labeled datasets is costly, time-intensive, and often impractical.Moreover, the subjective nature of emotions results in biased datasets, impacting the learning models' applicability in real-world scenarios. Motivated by how humans learn and conceptualize complex representations from an early age with minimal supervision, this approach demonstrates the effectiveness of leveraging prior experience to adapt to new situations. Unsupervised or self-supervised learning models draw inspiration from this paradigm. Initially, they aim to establish a general representation learning from unlabeled data, akin to the foundational prior experience in human learning. These representations should adhere to criteria like invariance, interpretability, and effectiveness. Subsequently, these learned representations are applied to downstream tasks with limited labeled data, such as emotion recognition. This mirrors the assimilation of new situations in human learning. In this thesis, we aim to propose unsupervised and self-supervised representation learning methods designed explicitly for multimodal and sequential data and to explore their potential advantages in the context of emotion recognition tasks. The main contributions of this thesis encompass:1. Developing generative models via unsupervised or self-supervised learning for audiovisual speech representation learning, incorporating joint temporal and multimodal (audiovisual) modeling.2. Structuring the latent space to enable disentangled representations, enhancing interpretability by controlling human-interpretable latent factors.3. Validating the effectiveness of our approaches through both qualitative and quantitative analyses, in particular on emotion recognition task. Our methods facilitate signal analysis, transformation, and generation
Freire, Marco. "Layout problems under topological constraints for computational fabrication." Electronic Thesis or Diss., Université de Lorraine, 2024. http://www.theses.fr/2024LORR0073.
Повний текст джерелаLayout problems appear in many areas of engineering and computer science. Typically, a layout problem requires to spatially arrange and interconnect a number of geometric elements in a domain. The elements can have a fixed or variable size, as well as an arbitrary shape. The domain may be be a volume, a planar region or a surface. It may be fixed or allowed to reshape. The interconnections may be simple paths, shared contact regions, or both. A set of constraints and objectives complement the problem definition, such as minimizing interconnection length, fixed positions for some elements, and many others. Layout problems are ubiquitous: floorplanning in architectural design, video game level design, industrial facility layout planning, electronics physical layout design, and so on. Topological constraints often arise in layout problems. Topology considers objects as defined by their elements' neighborhoods, without consideration for their specific geometry of placement. For example, a graph is a purely topological structure, consisting only of the relationships between its nodes. On the other hand, a graph drawing needs to specify the position of its nodes, i.e. the geometry of the graph. This thesis focuses on tackling two specific layout problems subject to topological constraints arising in computational design and fabrication. These are electronic circuit physical layout generation and 3D printing support generation. The first contribution is an entire system for the design of freeform RGB LED displays through bendable circuit boards. Typical rigid PCBs are made to bend by strategically using kerfing, i.e. cutting patterns into the board to create `hinges' where it needs to fold. The system takes a low-poly mesh as an input and outputs fabrication-ready blueprints, that can be sent to any online PCB manufacturer. After fabrication, the display is obtained by folding the circuit over the 3D printed mesh. The LEDs are commonly found on commercially available LED strips and are easy to control. Thus, the display can be used through a programmable interface to generate impressive lighting effects in real time. The global layout problem is decomposed into local per-triangle sub-problems by exploiting the chain topology of the electronic circuit, the final layout being obtained by stitching the local solutions. Instead of traditionally following the physical design pipeline, i.e. schematics design, component placement and routing; we decide the number of components, their placement and their routing per-triangle on the fly. The second contribution is a procedural algorithm for generating bridges-and-pillars supports for 3D printing. These supports have been shown to print reliably and in a stable manner in [DHL14]. Unfortunately, the previous algorithm struggles to generate supports that do not intersect the object, leaving visible scars on its surface after support removal. Additionally, its complexity scales with the number of points to support. We propose an algorithm based on emph{Model Synthesis} (MS) [Mer09] to generate these supports, with an implicit knowledge of object avoidance and a complexity independent of the number of points to support. Our algorithm works on a voxelized representation of the object. The supports are encoded in the algorithm with a set of labels, each representing a part of the structure (e.g. a pillar block, a bridge block, a pillar-bridge junction); and a set of adjacency constraints defining all possible label combinations in every direction. The supports for an object are generated top to bottom by repeatedly assigning labels to voxels and propagating constraints to remove inconsistent labels in the domain. The algorithm, adjacency constraints and heuristics are co-designed to avoid the need for trial-and-error or backtracking, typical of MS and similar approaches
Côté, Marc-Alexandre. "Réseaux de neurones génératifs avec structure." Thèse, Université de Sherbrooke, 2017. http://hdl.handle.net/11143/10489.
Повний текст джерелаLe, Roy Christine. "De la génération hippie à la génération yuppie aux Etats-Unis." Paris 4, 1987. http://www.theses.fr/1987PA040152.
Повний текст джерелаThe hippies are the descendants of the beat generation of the fifties. Both groups showed non-conformism. The hippies tried to create a new type of family based on companionate marriage. They liked life in a community, close to nature. They were against war, but refused to take part in public demonstrations in the streets. They were nonviolent, above all. They rejected western materialism. Eastern philosophy was the only antidote for them. Heyday could only be reached with drugs. The songs of the sixties and seventies tried to awake the American consciousness. At the end of the seventies, the movement faded some hippies became artists, others became contestants and others yuppies. They are the present-day trends of the united-states. They are young urban professionals. They are serious and ambitious. They aspire to glory. They dream of power and money. They like sports and eat only healthy food. They have comfortable incomes. In politics, they are at the same time (in social matters) and conservative (for their business). They work harder than their parents to succeed, (the economical conjuncture is not particularly good for them. ) In the seventies, the American had lost the notion of American dream, after several disillusions: Vietnam, Watergate, energy crisis,. . . , but on the contrary, the yuppies think it is possible. If you really want it badly. Despite the diametrally opposite conceptions, some hippies went through their judgment again and became yuppies
Peytavie, Adrien. "Génération procédurale de Monde." Phd thesis, Université Claude Bernard - Lyon I, 2010. http://tel.archives-ouvertes.fr/tel-00841373.
Повний текст джерелаMarechal, Nicolas. "Génération de contenu graphique." Phd thesis, Université Claude Bernard - Lyon I, 2010. http://tel.archives-ouvertes.fr/tel-00843000.
Повний текст джерелаPellenard, Bertrand. "Génération de maillages quadrangulaires." Nice, 2012. http://www.theses.fr/2012NICE4112.
Повний текст джерелаThe goal of this thesis is to devise methods and algorithms for the automatic generation of isotropic and anisotropic quadrilateral meshes. The first contribution is a method for isotropic quadrangular meshing of 2D domains, with control upon both element sizing and orientation. At the heart of our algorithm is an optimization procedure which uses several successive steps to improve the mesh quality criteria for size, orientation and regularity. This methodology allows the simultaneous control of element sizing and orientation. The second contribution is an automatic method for computing an anisotropic rectangular metric on piecewise smooth surfaces approximated by triangle meshes. This metric is derived from a user-specified maximum tolerance error, and is expressed in angular deviation of the normal to the surface. It only depends on a single intuitive tolerance parameter and is shown particularly well suited to preserve boundaries and sharp features. The third contribution is a method for anisotropic polygonal remeshing of surfaces. The algorithm uses a greedy optimization procedure which adds, removes and relocates generators on the surface to satisfy two criteria related to partitioning and mesh conformity. These generators induce first a surface decomposition from their associated metrics during the optimization step, the a surface partitioning during the final meshing step. This methodology provides a good fit between the metric and the mesh elements during the optimization step
Bonnard, Jennifer. "Génération d'images 3D HDR." Thesis, Reims, 2015. http://www.theses.fr/2015REIMS014/document.
Повний текст джерелаHDR imaging and 3D imaging are two areas in which the simultaneous but separate development has been growing in recent years. On the one hand, HDR (High Dynamic Range) imaging allows to extend the dynamic range of traditionnal images called LDR (Low Dynamic Range). On the other hand, 3Dimaging offers immersion in the shown film with the feeling to be part of the acquired scene. Recently, these two areas have been combined to provide 3D HDR images or videos but few viable solutions existand none of them is available to the public. In this thesis, we propose a method to generate 3D HDR images for autostereoscopic displays by adapting a multi-viewpoints camera to several exposures acquisition.To do that, neutral density filters are fixed on the objectives of the camera. Then, pixel matchingis applied to aggregate pixels that represent the same point in the acquired scene. Finally, radiance is calculated for each pixel of the set of images by using a weighted average of LDR values. An additiona lstep is necessary because some pixels have wrong radiance. We proposed a method based on the color of adjacent pixels and two methods based on the correction of the disparity of those pixels. The first method is based on the disparity of pixels of the neighborhood and the second method on the disparity independently calculated on each color channel. This pipeline allows the generation of 3D HDR image son each viewpoint. A tone-mapping algorithm is then applied on each of these images. Their composition with filters corresponding to the autostereoscopic screen used allows the visualization of the generated 3DHDR image
Rouxel-Labbé, Mael. "Génération de maillages anisotropes." Thesis, Université Côte d'Azur (ComUE), 2016. http://www.theses.fr/2016AZUR4150/document.
Повний текст джерелаIn this thesis, we study the generation of anisotropic meshes using the concepts of Delaunay triangulations and Voronoi diagrams. We first consider the framework of locally uniform anisotropic meshes introduced by Boissonnat, Wormser and Yvinec. Despite known theoretical guarantees, the practicality of this approach has only been hardly studied. An exhaustive empirical study is presented and reveals the strengths but also the overall impracticality of the method. In a second part, we investigate the anisotropic Voronoi diagram introduced by Labelle and Shewchuk and give conditions on a set of seeds such that the corresponding diagram has a dual that is an embedded triangulation in any dimension; an algorithm to generate such sets is devised. Using the same diagram, we propose an algorithm to generate efficiently anisotropic triangulations of low-dimensional manifolds embedded in high-dimensional spaces. Our algorithm is provable, but produces disappointing results. Finally, we study Riemannian Voronoi diagrams and introduce discrete Riemannian Voronoi diagrams, which employ recent developments in the numerical computation of geodesic distances and whose computation is accelerated through the use of an underlying anisotropic graph structure. We give conditions that guarantee that our discrete structure is combinatorially equivalent to the Riemannian Voronoi diagram and that its dual is an embedded triangulation, using both straight and curved simplices. We obtain significantly better results than with our other methods, but the overall utility of
Webanck, Antoine. "Génération procédurale d'effets atmosphériques." Thesis, Lyon, 2019. http://www.theses.fr/2019LYSE1109/document.
Повний текст джерелаThis thesis focusses on the synthetization of natural landscapes, and more particularly on their celestial part. The aspect of the sky is governed by plenty of atmospheric phenomena, among which clouds play a major role for they are recurrent and widespread. Even without directly considering the sky, the density of the clouds allows them to intensely modify the global illumination of a landscape. The work of this thesis thus focuses mainly on the editing, modelling and animation of cloud areas of landscape dimensions.Because the thermodynamic simulation of cloud formation is hard to control and its maximum resolution quickly limits the details of the simulated volume, we propose instead a procedural generation method. We build a lightweight cloudscape model as a hierarchy of functions. The finest details are obtained by composing procedural noises and reproduce the specific shapes of different kinds of clouds. The large-scale cloud presence is described at a high level and at different times by maps drawn by the user. These discrete maps are transformed into implicit static primitives and then interpolated by morphing, accounting for relief and winds in order to produce coherent trajectories. The implicit field obtained by mixing the interpolating primitives represents the spatiotemporal field of cloud density. Images are finally synthesized by rendering of the atmospheric participative medium according to our own implementation, executed in parallel on a graphic card
Abrishami, Homeira. "Génération de motifs persans." Paris 8, 2010. http://octaviana.fr/document/156463229#?c=0&m=0&s=0&cv=0.
Повний текст джерелаThis research offers an exhaustive classification of Persian Islamic ornamental patterns, based on a combination and a comparison of previous classifications, using historical Persian documents and decisive European scholarship. This research is unique because of the importance which we accord to Persian patterns, particularly to the Pre-Islamic ones that influenced later Islamic motifs. We then emphasize more recent Persian patterns and techniques (boteh, khataï et girih), whose development was local but influenced the entire Islamic world. Using many illustrations and examples, we describe these patterns and explain the process of their creation, highlighting the role that Greek and Persian mathematics played for centuries. We then analyse the importance of crystallography, from the XIX to XX centuries, first by linking symmetries of crystals (2, 3, 4 and 6 folds) with the periodic tilings of geometric ornaments, then by assimilating quasi-periodicity (5 fold symmetry) and Penrose’s tilings with the traditional technique of girih. Finally, we survey automatic computer generation of ornamental patterns
Fleury, Charles. "Génération lyrique et génération X : parcours de jeunesse de deux générations au Québec." Thesis, Université Laval, 2008. http://www.theses.ulaval.ca/2008/25107/25107.pdf.
Повний текст джерелаCordier, Nicolas. "Approches multi-atlas fondées sur l'appariement de blocs de voxels pour la segmentation et la synthèse d'images par résonance magnétique de tumeurs cérébrales." Thesis, Nice, 2015. http://www.theses.fr/2015NICE4111/document.
Повний текст джерелаThis thesis focuses on the development of automatic methods for the segmentation and synthesis of brain tumor Magnetic Resonance images. The main clinical perspective of glioma segmentation is growth velocity monitoring for patient therapy management. To this end, the thesis builds on the formalization of multi-atlas patch-based segmentation with probabilistic graphical models. A probabilistic model first extends classical multi-atlas approaches used for the segmentation of healthy brains structures to the automatic segmentation of pathological cerebral regions. An approximation of the marginalization step replaces the concept of local search windows with a stratification with respect to both atlases and labels. A glioma detection model based on a spatially-varying prior and patch pre-selection criteria are introduced to obtain competitive running times despite patch matching being non local. This work is validated and compared to state-of-the-art algorithms on publicly available datasets. A second probabilistic model mirrors the segmentation model in order to synthesize realistic MRI of pathological cases, based on a single label map. A heuristic method allows to solve for the maximum a posteriori and to estimate uncertainty of the image synthesis model. Iterating patch matching reinforces the spatial coherence of synthetic images. The realism of our synthetic images is assessed against real MRI, and against outputs of the state-of-the-art method. The junction of a tumor growth model to the proposed synthesis approach allows to generate databases of annotated synthetic cases
Gallas, Mohamed-Anis. "De l'intention à la solution architecturale : proposition d'une méthode d'assistance à la prise en compte de la lumière naturelle durant les phases amont de conception." Electronic Thesis or Diss., Université de Lorraine, 2013. http://www.theses.fr/2013LORR0101.
Повний текст джерелаThe daylight illuminates the architectural space and creates the necessary conditions to accommodate human activities. The illuminating function of daylight is joined to another sensitive dimension that attributes a distinctive identity to the designed space. The control of daylight in architectural environment needs some detailed and precise characteristics about aperture and walls features. The multiplicity of these characteristics is faced with the inaccuracy and the uncertainty of the design information available during the early steps. Our research aims to propose a design support method that takes into account the early design step features and helps designers to integrate there daylight atmosphere intentions in project. We propose a design support method structured as a declarative modelling process. The declarative process helps the designer to declare his daylight intentions and to translate them to potential solutions that could be integrated in his project. This method considers the designer intentions as the main design information used to help designer during the early design steps. The proposed method provides functionalities that could operate the inaccuracy and the uncertainty of the ideas research and formalisation steps. Our design support method was implemented in a prototype design support tool. The cognitive contribution and the ability to support the design activities of the proposed tool was evaluated and analysed in an experimental design context
Lanusse, Patrick. "De la commande CRONE de première génération à la commande CRONE de troisième génération." Bordeaux 1, 1994. http://www.theses.fr/1994BOR10543.
Повний текст джерелаMognol, Pascal. "Contribution à la génération automatique de gammes en tournage : génération dirigée par évaluation progressive." Cachan, Ecole normale supérieure, 1994. http://www.theses.fr/1994DENS0019.
Повний текст джерелаParigot, Didier. "Contribution à la programmation générative." Habilitation à diriger des recherches, Université de Nice Sophia-Antipolis, 2003. http://tel.archives-ouvertes.fr/tel-00506070.
Повний текст джерелаMairesse, Yann. "Génération et caractérisation d'impulsions attosecondes." Phd thesis, Université Paris Sud - Paris XI, 2005. http://tel.archives-ouvertes.fr/tel-00011620.
Повний текст джерелаEn transposant une technique d'interférométrie spectrale couramment utilisée pour la caractérisation complète d'impulsions infrarouges (SPIDER), nous effectuons une caractérisation complète monocoup du profil temporel d'harmoniques individuelles, à l'échelle femtoseconde.
Ensuite, nous étudions expérimentalement la structure attoseconde du rayonnement harmonique, et mettons en évidence une dérive temporelle dans l'émission : les harmoniques les plus faibles sont émises avant les plus élevées. Cette dérive, qui est directement liée à la dynamique électronique microscopique dans le processus de génération, limite la durée d'impulsion que l'on peut obtenir en augmentant la largeur spectrale. Nous présentons les résultats de l'optimisation des conditions de génération afin d'améliorer la synchronisation dans l'émission. Nous montrons également la possibilité de recomprimer les impulsions attosecondes.
Enfin, nous proposons une nouvelle technique pour la caractérisation complète d'impulsions attosecondes arbitraires : FROGCRAB. Elle permettrait une mesure simultanée des caractéristiques femtoseconde et attoseconde du rayonnement, et ainsi une connaissance complète de la source lumineuse attoseconde en vue de son utilisation dans des expériences d'applications.
Villeneuve, D. "Logiciel de génération de colonnes." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 2000. http://www.collectionscanada.ca/obj/s4/f2/dsk2/ftp02/NQ53547.pdf.
Повний текст джерелаPetitjean, Simon. "Génération modulaire de grammaires formelles." Thesis, Orléans, 2014. http://www.theses.fr/2014ORLE2048/document.
Повний текст джерелаThe work presented in this thesis aim at facilitating the development of resources for natural language processing. Resources of this type take different forms, because of the existence of several levels of linguistic description (syntax, morphology, semantics, . . . ) and of several formalisms proposed for the description of natural languages at each one of these levels. The formalisms featuring different types of structures, a unique description language is not enough: it is necessary to create a domain specific language (or DSL) for every formalism, and to implement a new tool which uses this language, which is a long a complex task. For this reason, we propose in this thesis a method to assemble in a modular way development frameworks specific to tasks of linguistic resource generation. The frameworks assembled thanks to our method are based on the fundamental concepts of the XMG (eXtensible MetaGrammar) approach, allowing the generation of tree based grammars. The method is based on the assembling of a description language from reusable bricks, and according to a unique specification file. The totality of the processing chain for the DSL is automatically assembled thanks to the same specification. In a first time, we validated this approach by recreating the XMG tool from elementary bricks. Some collaborations with linguists also brought us to assemble compilers allowing the description of morphology and semantics
Schmitt, Maxime. "Génération automatique de codes adaptatifs." Thesis, Strasbourg, 2019. http://www.theses.fr/2019STRAD029.
Повний текст джерелаIn this thesis we introduce a new application programming interface to help developers to optimize an application with approximate computing techniques. This interface is provided as a language extension to advise the compiler about the parts of the program that may be optimized with approximate computing and what can be done about them. The code transformations of the targeted regions are entirely handled by the compiler to produce an adaptive software. The produced adaptive application allocates more computing power to the locations where more precision is required, and may use approximations where the precision is secondary. We automate the discovery of the optimization parameters for the special class of stencil programs which are common in signal/image processing and numerical simulations. Finally, we explore the possibility of compressing the application data using the wavelet transform and we use information found in this basis to locate the areas where more precision may be needed
Thiéblin, Elodie. "Génération automatique d'alignements complexes d'ontologies." Thesis, Toulouse 3, 2019. http://www.theses.fr/2019TOU30135.
Повний текст джерелаThe Linked Open Data (LOD) cloud is composed of data repositories. The data in the repositories are described by vocabularies also called ontologies. Each ontology has its own terminology and model. This leads to heterogeneity between them. To make the ontologies and the data they describe interoperable, ontology alignments establish correspondences, or links between their entities. There are many ontology matching systems which generate simple alignments, i.e., they link an entity to another. However, to overcome the ontology heterogeneity, more expressive correspondences are sometimes needed. Finding this kind of correspondence is a fastidious task that can be automated. In this thesis, an automatic complex matching approach based on a user's knowledge needs and common instances is proposed. The complex alignment field is still growing and little work address the evaluation of such alignments. To palliate this lack, we propose an automatic complex alignment evaluation system. This system is based on instances. A famous alignment evaluation dataset has been extended for this evaluation
Jeyasothy, Adulam. "Génération d'explications post-hoc personnalisées." Electronic Thesis or Diss., Sorbonne université, 2024. http://www.theses.fr/2024SORUS027.
Повний текст джерелаThis thesis is in the field of eXplainable AI (XAI). We focus on post-hoc interpretability methods that aim to explain to a user the prediction for a specific data made by a trained decision model. To increase the interpretability of explanations, this thesis studies the integration of user knowledge into these methods, and thus aims to improve the understandability of the explanation by generating personalized explanations tailored to each user. To this end, we propose a general formalism that explicitly integrates knowledge via a new criterion in the interpretability objectives. This formalism is then declined for different types of knowledge and different types of explanations, particularly counterfactual examples, leading to the proposal of several algorithms (KICE, Knowledge Integration in Counterfactual Explanation, rKICE for its variant including knowledge expressed by rules and KISM, Knowledge Integration in Surrogate Models). The issue of aggregating classical quality and knowledge compatibility constraints is also studied, and we propose to use Gödel's integral as an aggregation operator. Finally, we discuss the difficulty of generating a single explanation suitable for all types of users and the notion of diversity in explanations
Flandrin, Nicolas. "Génération de maillage hybride pour les simulateurs de réservoir pétrolier de nouvelle génération : extension 3D." Troyes, 2005. http://www.theses.fr/2005TROY0009.
Повний текст джерелаDuring the exploitation of an oil reservoir, it is important to predict the recovery of hydrocarbons and to optimize its production. A better comprehension of the physical phenomena requires to simulate 3D multiphase flows in increasingly complex geological structures. In this thesis, we are interested in this spatial discretization and we propose to extend in 3D the 2D hybrid model proposed by IFP in 1998 where the radial characteristics of the flows in the vicinity of wells are directly taken into account in the geometry. In these hybrid meshes, the wells and their drainage area are described by structured radial circular meshes and the reservoirs are represented by structured meshes that can be non uniform Cartesian or Corner Point Geometry grids. In order to generate a global conforming mesh, unstructured transition meshes based on power diagrams and satisfying finite volume properties are used to connect together the structured meshes. Two methods have been implemented to generate these transition meshes: a method using a Delaunay triangulation and another one using a frontal approach. In addition, some criteria are introduced to measure the quality or the transition meshes and optimization procedures are proposed to increase some of these criteria under finite volume properties constraints
Lucas, Thomas. "Modèles génératifs profonds : sur-généralisation et abandon de mode." Thesis, Université Grenoble Alpes, 2020. http://www.theses.fr/2020GRALM049.
Повний текст джерелаThis dissertation explores the topic of generative modelling of natural images,which is the task of fitting a data generating distribution.Such models can be used to generate artificial data resembling the true data, or to compress images.Latent variable models, which are at the core of our contributions, seek to capture the main factors of variations of an image into a variable that can be manipulated.In particular we build on two successful latent variable generative models, the generative adversarial network (GAN) and Variational autoencoder (VAE) models.Recently GANs significantly improved the quality of images generated by deep models, obtaining very compelling samples.Unfortunately these models struggle to capture all the modes of the original distribution, ie they do not cover the full variability of the dataset.Conversely, likelihood based models such as VAEs typically cover the full variety of the data well and provide an objective measure of coverage.However these models produce samples of inferior visual quality that are more easily distinguished from real ones.The work presented in this thesis strives for the best of both worlds: to obtain compelling samples while modelling the full support of the distribution.To achieve that, we focus on i) the optimisation problems used and ii) practical model limitations that hinder performance.The first contribution of this manuscript is a deep generative model that encodes global image structure into latent variables, built on the VAE, and autoregressively models low level detail.We propose a training procedure relying on an auxiliary loss function to control what information is captured by the latent variables and what information is left to an autoregressive decoder.Unlike previous approaches to such hybrid models, ours does not need to restrict the capacity of the autoregressive decoder to prevent degenerate models that ignore the latent variables.The second contribution builds on the standard GAN model, which trains a discriminator network to provide feedback to a generative network.The discriminator usually assesses the quality of individual samples, which makes it hard to evaluate the variability of the data.Instead we propose to feed the discriminator with emph{batches} that mix both true and fake samples, and train it to predict the ratio of true samples in the batch.These batches work as approximations of the distribution of generated images and allows the discriminator to approximate distributional statistics.We introduce an architecture that is well suited to solve this problem efficiently,and show experimentally that our approach reduces mode collapse in GANs on two synthetic datasets, and obtains good results on the CIFAR10 and CelebA datasets.The mutual shortcomings of VAEs and GANs can in principle be addressed by training hybrid models that use both types of objective.In our third contribution, we show that usual parametric assumptions made in VAEs induce a conflict between them, leading to lackluster performance of hybrid models.We propose a solution based on deep invertible transformations, that trains a feature space in which usual assumptions can be made without harm.Our approach provides likelihood computations in image space while being able to take advantage of adversarial training.It obtains GAN-like samples that are competitive with fully adversarial models while improving likelihood scores over existing hybrid models at the time of publication, which is a significant advancement
Dangauthier, Pierre-Charles. "Fondations, méthode et applications de l'apprentissage bayésien." Phd thesis, Grenoble INPG, 2007. http://tel.archives-ouvertes.fr/tel-00267643.
Повний текст джерелаChambefort, Françoise. "Mimèsis du flux, exploration des potentialités narratives des flux de données." Thesis, Bourgogne Franche-Comté, 2020. http://www.theses.fr/2020UBFCC004.
Повний текст джерелаSometimes called stream art or data art, digital art seizes data streams as its raw materials. Choosing a path of creative research, this thesis explores the story-telling potentialities of data streams. Structured around technical, social, semiotic and aesthetic approaches, its thinking draws on various fields of study : information and communication sciences, but also computer sciences, cognitive sciences, philosophy, sociology and narratology. The work Lucette, Gare de Clichy was especially designed to answer the researched question. The conformation of the work allowed for two different versions of it : a screen version and a performance. It is studied in all its stages, from its creation process to the public's response to it. Jonathan Fletcher Moore's installation, Artificial Killing Machine, is also analyzed. First, our object of research - stories made from a real-time data stream - is defined and the concept of data mills is crafted to refer to this type of work. Then four hypothesis are formulated and individually verified. If data mills are to be able to form a narrative representation, they must free themselves from the logic of action. Thus can fiction become powered by reality. The metaphor that links the data originated in reality and the crafted fiction generates in the viewer a shifting of focus between what is compared and what compares. This switching-metaphor has the power to reinforce the meaning it carries. Data mills are therefore able to convey the contingency of life as experienced by a vulnerable individual, tossed back and forth between objective and subjective time
Mehr, Éloi. "Unsupervised Learning of 3D Shape Spaces for 3D Modeling." Electronic Thesis or Diss., Sorbonne université, 2019. http://www.theses.fr/2019SORUS566.
Повний текст джерелаEven though 3D data is becoming increasingly more popular, especially with the democratization of virtual and augmented experiences, it remains very difficult to manipulate a 3D shape, even for designers or experts. Given a database containing 3D instances of one or several categories of objects, we want to learn the manifold of plausible shapes in order to develop new intelligent 3D modeling and editing tools. However, this manifold is often much more complex compared to the 2D domain. Indeed, 3D surfaces can be represented using various embeddings, and may also exhibit different alignments and topologies. In this thesis we study the manifold of plausible shapes in the light of the aforementioned challenges, by deepening three different points of view. First of all, we consider the manifold as a quotient space, in order to learn the shapes’ intrinsic geometry from a dataset where the 3D models are not co-aligned. Then, we assume that the manifold is disconnected, which leads to a new deep learning model that is able to automatically cluster and learn the shapes according to their typology. Finally, we study the conversion of an unstructured 3D input to an exact geometry, represented as a structured tree of continuous solid primitives
Bertault, Francois. "Génération et tracé de structures décomposables." Phd thesis, Université Henri Poincaré - Nancy I, 1997. http://tel.archives-ouvertes.fr/tel-00557784.
Повний текст джерела