Tesi sul tema "Optimisation et apprentissage distribués"
Cita una fonte nei formati APA, MLA, Chicago, Harvard e in molti altri stili
Vedi i top-50 saggi (tesi di laurea o di dottorato) per l'attività di ricerca sul tema "Optimisation et apprentissage distribués".
Accanto a ogni fonte nell'elenco di riferimenti c'è un pulsante "Aggiungi alla bibliografia". Premilo e genereremo automaticamente la citazione bibliografica dell'opera scelta nello stile citazionale di cui hai bisogno: APA, MLA, Harvard, Chicago, Vancouver ecc.
Puoi anche scaricare il testo completo della pubblicazione scientifica nel formato .pdf e leggere online l'abstract (il sommario) dell'opera se è presente nei metadati.
Vedi le tesi di molte aree scientifiche e compila una bibliografia corretta.
Martinez, Medina Lourdes. "Optimisation des requêtes distribuées par apprentissage". Thesis, Grenoble, 2014. http://www.theses.fr/2014GRENM015.
Testo completoDistributed data systems are becoming increasingly complex. They interconnect devices (e.g. smartphones, tablets, etc.) that are heterogeneous, autonomous, either static or mobile, and with physical limitations. Such devices run applications (e.g. virtual games, social networks, etc.) for the online interaction of users producing / consuming data on demand or continuously. The characteristics of these systems add new dimensions to the query optimization problem, such as multi-optimization criteria, scarce information on data, lack of global system view, among others. Traditional query optimization techniques focus on semi (or not at all) autonomous systems. They rely on information about data and make strong assumptions about the system behavior. Moreover, most of these techniques are centered on the optimization of execution time only. The difficulty for evaluating queries efficiently on nowadays applications motivates this work to revisit traditional query optimization techniques. This thesis faces these challenges by adapting the Case Based Reasoning (CBR) paradigm to query processing, providing a way to optimize queries when there is no prior knowledge of data. It focuses on optimizing queries using cases generated from the evaluation of similar past queries. A query case comprises: (i) the query, (ii) the query plan and (iii) the measures (computational resources consumed) of the query plan. The thesis also concerns the way the CBR process interacts with the query plan generation process. This process uses classical heuristics and makes decisions randomly (e.g. when there are no statistics for join ordering and selection of algorithms, routing protocols). It also (re)uses cases (existing query plans) for similar queries parts, improving the query optimization, and therefore evaluation efficiency. The propositions of this thesis have been validated within the CoBRa optimizer developed in the context of the UBIQUEST project
Jankee, Christopher. "Optimisation par métaheuristique adaptative distribuée en environnement de calcul parallèle". Thesis, Littoral, 2018. http://www.theses.fr/2018DUNK0480/document.
Testo completoTo solve discrete optimization problems of black box type, many stochastic algorithms such as evolutionary algorithms or metaheuristics exist and prove to be particularly effective according to the problem to be solved. Depending on the observed properties of the problem, choosing the most relevant algorithm is a difficult problem. In the original framework of parallel and distributed computing environments, we propose and analyze different adaptive optimization algorithm selection strategies. These selection strategies are based on reinforcement learning methods automatic, from the field of artificial intelligence, and on information sharing between computing nodes. We compare and analyze selection strategies in different situations. Two types of synchronous distributed computing environment are discussed : the island model and the master-slave model. On the set of nodes synchronously at each iteration, the adaptive selection strategy chooses an algorithm according to the state of the search for the solution. In the first part, two problems OneMax and NK, one unimodal and the other multimodal, are used as benchmarks for this work. Then, to better understand and improve the design of adaptive selection strategies, we propose a modeling of the optimization problem and its local search operator. In this modeling, an important characteristic is the average gain of an operator according to the fitness of the candidate solution. The model is used in the synchronous framework of the master-slave model. A selection strategy is broken down into three main components : the aggregation of the rewards exchanged, the learning scheme and the distribution of the algorithms on the computing nodes. In the final part, we study three scenarios, and we give keys to understanding the relevant use of adaptive selection strategies over naïve strategies. In the framework of the master-slave model, we study the different ways of aggregating the rewards on the master node, the distribution of the optimization algorithms of the nodes of computation and the time of communication. This thesis ends with perspectives in the field of distributed adaptive stochastic optimization
Mhanna, Elissa. "Beyond gradients : zero-order approaches to optimization and learning in multi-agent environments". Electronic Thesis or Diss., université Paris-Saclay, 2024. http://www.theses.fr/2024UPASG123.
Testo completoThe rise of connected devices and the data they produce has driven the development of large-scale applications. These devices form distributed networks with decentralized data processing. As the number of devices grows, challenges like communication overhead and computational costs increase, requiring optimization methods that work under strict resource constraints, especially where derivatives are unavailable or costly. This thesis focuses on zero-order (ZO) optimization methods are ideal for scenarios where explicit function derivatives are inaccessible. ZO methods estimate gradients based only on function evaluations, making them highly suitable for distributed and federated learning environments where devices collaborate to solve global optimization tasks with limited information and noisy data. In the first chapter, we address distributed ZO optimization for strongly convex functions across multiple agents in a network. We propose a distributed zero-order projected gradient descent algorithm that uses one-point gradient estimates, where the function is queried only once per stochastic realization, and noisy function evaluations estimate the gradient. The chapter establishes the almost sure convergence of the algorithm and derives theoretical upper bounds on the convergence rate. With constant step sizes, the algorithm achieves a linear convergence rate. This is the first time this rate has been established for one-point (and even two-point) gradient estimates. We also analyze the effects of diminishing step sizes, establishing a convergence rate that matches centralized ZO methods' lower bounds. The second chapter addresses the challenges of federated learning (FL) which is often hindered by the communication bottleneck—the high cost of transmitting large amounts of data over limited-bandwidth networks. To address this, we propose a novel zero-order federated learning (ZOFL) algorithm that reduces communication overhead using one-point gradient estimates. Devices transmit scalar values instead of large gradient vectors, lowering the data sent over the network. Moreover, the algorithm incorporates wireless communication disturbances directly into the optimization process, eliminating the need for explicit knowledge of the channel state. This approach is the first to integrate wireless channel properties into a learning algorithm, making it resilient to real-world communication issues. We prove the almost sure convergence of this method in nonconvex optimization settings, establish its convergence rate, and validate its effectiveness through experiments. In the final chapter, we extend the ZOFL algorithm to include two-point gradient estimates. Unlike one-point estimates, which rely on a single function evaluation, two-point estimates query the function twice, providing a more accurate gradient approximation and enhancing the convergence rate. This method maintains the communication efficiency of one-point estimates, where only scalar values are transmitted, and relaxes the assumption that the objective function must be bounded. The chapter demonstrates that the proposed two-point ZO method achieves linear convergence rates for strongly convex and smooth objective functions. For nonconvex problems, the method shows improved convergence speed, particularly when the objective function is smooth and K-gradient-dominated, where a linear rate is also achieved. We also analyze the impact of constant versus diminishing step sizes and provide numerical results showing the method's communication efficiency compared to other federated learning techniques
Vicard, Annie. "Formalisation et optimisation des systèmes informatiques distribués temps réel embarqués". Paris 13, 1999. http://www.theses.fr/1999PA132032.
Testo completoMériaux, François. "Théorie des jeux et apprentissage pour les réseaux sans fil distribués". Phd thesis, Université Paris Sud - Paris XI, 2013. http://tel.archives-ouvertes.fr/tel-00952069.
Testo completoZerrik, El Hassan. "Controlabilité et observalité régionales d'une classe de systèmes distribués". Perpignan, 1994. http://www.theses.fr/1994PERP0176.
Testo completoVan, Grieken Milagros. "Optimisation pour l'apprentissage et apprentissage pour l'optimisation". Phd thesis, Université Paul Sabatier - Toulouse III, 2004. http://tel.archives-ouvertes.fr/tel-00010106.
Testo completoBERNY, ARNAUD. "Apprentissage et optimisation statistiques. Application a la radiotelephonie mobile". Nantes, 2000. http://www.theses.fr/2000NANT2081.
Testo completoLe, Lann Marie-Véronique. "Commande prédictive et commande par apprentissage : étude d'une unité pilote d'extraction, optimisation par apprentissage". Toulouse, INPT, 1988. http://www.theses.fr/1988INPT023G.
Testo completoLe, Lann Marie-Véronique. "Commande prédictive et commande par apprentissage étude d'une unité pilote d'extraction, optimisation par apprentissage /". Grenoble 2 : ANRT, 1988. http://catalogue.bnf.fr/ark:/12148/cb37615168p.
Testo completoBertrand, Hadrien. "Optimisation d'hyper-paramètres en apprentissage profond et apprentissage par transfert : applications en imagerie médicale". Thesis, Université Paris-Saclay (ComUE), 2019. http://www.theses.fr/2019SACLT001/document.
Testo completoIn the last few years, deep learning has changed irrevocably the field of computer vision. Faster, giving better results, and requiring a lower degree of expertise to use than traditional computer vision methods, deep learning has become ubiquitous in every imaging application. This includes medical imaging applications. At the beginning of this thesis, there was still a strong lack of tools and understanding of how to build efficient neural networks for specific tasks. Thus this thesis first focused on the topic of hyper-parameter optimization for deep neural networks, i.e. methods for automatically finding efficient neural networks on specific tasks. The thesis includes a comparison of different methods, a performance improvement of one of these methods, Bayesian optimization, and the proposal of a new method of hyper-parameter optimization by combining two existing methods: Bayesian optimization and Hyperband.From there, we used these methods for medical imaging applications such as the classification of field-of-view in MRI, and the segmentation of the kidney in 3D ultrasound images across two populations of patients. This last task required the development of a new transfer learning method based on the modification of the source network by adding new geometric and intensity transformation layers.Finally this thesis loops back to older computer vision methods, and we propose a new segmentation algorithm combining template deformation and deep learning. We show how to use a neural network to predict global and local transformations without requiring the ground-truth of these transformations. The method is validated on the task of kidney segmentation in 3D US images
Bertrand, Hadrien. "Optimisation d'hyper-paramètres en apprentissage profond et apprentissage par transfert : applications en imagerie médicale". Electronic Thesis or Diss., Université Paris-Saclay (ComUE), 2019. http://www.theses.fr/2019SACLT001.
Testo completoIn the last few years, deep learning has changed irrevocably the field of computer vision. Faster, giving better results, and requiring a lower degree of expertise to use than traditional computer vision methods, deep learning has become ubiquitous in every imaging application. This includes medical imaging applications. At the beginning of this thesis, there was still a strong lack of tools and understanding of how to build efficient neural networks for specific tasks. Thus this thesis first focused on the topic of hyper-parameter optimization for deep neural networks, i.e. methods for automatically finding efficient neural networks on specific tasks. The thesis includes a comparison of different methods, a performance improvement of one of these methods, Bayesian optimization, and the proposal of a new method of hyper-parameter optimization by combining two existing methods: Bayesian optimization and Hyperband.From there, we used these methods for medical imaging applications such as the classification of field-of-view in MRI, and the segmentation of the kidney in 3D ultrasound images across two populations of patients. This last task required the development of a new transfer learning method based on the modification of the source network by adding new geometric and intensity transformation layers.Finally this thesis loops back to older computer vision methods, and we propose a new segmentation algorithm combining template deformation and deep learning. We show how to use a neural network to predict global and local transformations without requiring the ground-truth of these transformations. The method is validated on the task of kidney segmentation in 3D US images
Henriet, Julien. "Evaluation, optimisation et validation de protocoles distribués de gestion de la concurrence pour les interactions coopératives". Besançon, 2005. http://www.theses.fr/2005BESA2023.
Testo completoCooperative work over Internet introduces constraints in terms of access and modification of shared objects. Users need to access the same objects concurrently in real-time. At each moment, the same image of the production area is to be shared on each connected site. The first chapter of this thesis is a state of the art of communication protocols, consistency management protocols and telemedicine platforms that allow collaborative work. The second chapter presents an new protocol for consistency management over that kind of platform. A probabilistic study of this new protocol proves and evaluates the optimization and the cost of this new protocol called the Optimistic Pilgrim. An analysis, an optimization and a validation of the Chameleon protocol dedicated to communication management over a collaborative platform is presented in the third chapter. At last, the fourth chapter evaluates the performance of these protocols through an implementation of a prototype and also presents the adequation of each protocol to each tool of the collaborative teleneurological platform of the TéNeCi project
Larrousse, Benjamin. "Structure d’information, stratégies de communication et application aux réseaux distribués". Thesis, Paris 11, 2014. http://www.theses.fr/2014PA112373/document.
Testo completoThis thesis studies distributed optimization problems with different observation structuresand application to wireless network and Smart Grids problems. Specifically, an asymmetricobservation structure between two agents is considered, where a first agent has full knowledgeabout the realization of a random state, and the other agent does not know anything about thisstate. In this context, the question is how to transmit information from the first agent to thesecond agent in order to use in an optimal way the communication resources. Several modelsare studied in this thesis. For all of them, a common element is that the information source hasto be encoded in an appropriate manner to optimize the use of the system’s configuration. Afirst model is studied where no dedicated channel for communication is available between agentsand they have the same objective function. Therefore, the only way communication is possible isthrough the actions chosen by agents. As actions are payoff relevant, the first agent has to findthe optimal tradeoff between transmission of information and payoff maximization. The informedagent encodes his knowledge about the state into his actions, which will be imperfectly observedby the second agent. The latter will decode the information and choose his actions in order tomaximize the common objective function. We use tools from information theory to characterizethis optimal tradeoff by an information constraint, and apply this scenario to a power controlproblem in an interference channel setting. Our new strategy (the coded power control ) givessome promising gains compare to classical approaches.In a second part, we consider that there exists a dedicated channel for communication, that isto say the actions of the informed agent are not payoff relevant and are only useful for transmissionof information. Furthermore, agents are supposed to have diverging interests, so that the informedagent does not necessarily have an incentive to send all his knowledge to the uninformed agent.Game theory and Cheap talk game in particular appears to be the right framework to analyzethis problem. We characterize the signal scheme that agents will agree on. This scheme willlead to a Nash Equilibrium, thus will optimize the way communication is done. This model is ofparticular interest for electrical vehicles networks where an electrical vehicle has to send his needin term of power to an aggregator which will choose an effective charging level for the electricalvehicle. The latter only cares about his need in term of power whereas the aggregator also takesinto account the network status. The considered model help to optimize the way the network isused.We finally consider a model with more than two agents, where the main goal is for all agentsto retrieve perfect observations of all past actions of all agents. This is of particular interest ina game theory point of view to characterize the long term expected utilities of the agents. Inthis model, we add an encoder who perfectly oberves all past actions and will help agents tohave perfect monitoring. In fact, this is possible if the right information constraint is satisfied.We thus characterized the latter, using a hybrid coding scheme combining classical informationtheoretic scheme and tools from graph theory
Zennir, Youcef. "Apprentissage par renforcement et systèmes distribués : application à l'apprentissage de la marche d'un robot hexapode". Lyon, INSA, 2004. http://theses.insa-lyon.fr/publication/2004ISAL0034/these.pdf.
Testo completoThe goal of this thesis is to study and to develop reinforcement learning techniques in order a hexapod robot to learn to walk. The main assumption on which this work is based is that effective gaits can be obtained as the control of the movements is distributed on each leg rather than centralised in a single decision centre. A distributed approach of the Q-learning technique is adopted in which the agents contributing to the same global objective perform their own learning process taking into account or not the other agents. The centralised and distributed approaches are compared. Different simulations and tests are carried out so as to generate stable periodic gaits. The influence of the learning parameters on the quality of the gaits are studied. The walk appears as an emerging phenomenon from the individual movements of the legs. Problems of fault tolerance and lack of state information are investigated. Finally it is verified that with the developed algorithm the simulated robot learns how to reach a desired trajectory while controlling its posture
Canon, Louis-claude. "Outils et algorithmes pour gérer l'incertitude lors de l'ordonnancement d'application sur plateformes distribuées". Thesis, Nancy 1, 2010. http://www.theses.fr/2010NAN10097/document.
Testo completoThis thesis consists in revisiting traditional scheduling problematics in computational environments, and considering the adjunction of uncertainty in the models. We adopt here a wide definition of uncertainty that encompasses the intrinsic stochastic nature of some phenomena (e.g., processor failures that follow a Poissonian distribution) and the imperfection of model characteristics (e.g., inaccuracy of the costs in a model due to a bias in measurements). We also consider uncertainties that stem from indeterminations such as the user behaviors that are uncontrolled although being deterministic. Scheduling, in its general form, is the operation that assigns requests to resources in some specific way. In distributed environments, we are concerned by a workload (i.e., a set of tasks) that needs to be executed onto a computational platform (i.e., a set of processors). Therefore, our objective is to specify how tasks are mapped onto processors. Produced schedules can be evaluated through many different metrics (e.g., processing time of the workload, resource usage, etc) and finding an optimal schedule relatively to some metric constitutes a challenging issue. Probabilistic tools and multi-objectives optimization techniques are first proposed for tackling new metrics that arise from the uncertainty. In a second part, we study several uncertainty-related criteria such as the robustness (stability in presence of input variations) or the reliability (probability of success) of a schedule
Pasquero, Oudomsack Pierre. "Optimisation de systèmes de télévision numérique terrestre : Estimation de canal, Synchronisation et Schémas multi-antennes distribués". Phd thesis, INSA de Rennes, 2011. http://tel.archives-ouvertes.fr/tel-00635282.
Testo completoKebbal, Djemai. "Tolérance aux fautes et ordonnancement adaptatif dans les systèmes distribués hétérogènes". Lille 1, 2000. https://pepite-depot.univ-lille.fr/LIBRE/Th_Num/2000/50376-2000-316.pdf.
Testo completoPhilip, Julien. "Édition et rendu à base d’images multi-vues par apprentissage profond et optimisation". Thesis, Université Côte d'Azur, 2020. http://www.theses.fr/2020COAZ4048.
Testo completoComputer-generated imagery (CGI) takes a growing place in our everyday environment. Whether it is in video games or movies, CGI techniques are constantly improving in quality but also require ever more qualitative artistic content which takes a growing time to create. With the emergence of virtual and augmented reality, often comes the need to render or re-render assets that exist in our world. To allow widespread use of CGI in applications such as telepresence or virtual visits, the need for manual artistic replication of assets must be removed from the process. This can be done with the help of Image-Based Rendering (IBR) techniques that allow scenes or objects to be rendered in a free-viewpoint manner from a set of sparse input photographs. While this process requires little to no artistic work, it also does not allow for artistic control or editing of scene content. In this dissertation, we explore Multi-view Image Editing and Rendering. To allow casually captured scenes to be rendered with content alterations such as object removal, lighting edition, or scene compositing, we leverage the use of optimization techniques and modern deep-learning. We design our methods to take advantage of all the information present in multi-view content while handling specific constraints such as multi-view coherency. For object removal, we introduce a new plane-based multi-view inpainting algorithm. Planes are a simple yet effective way to fill geometry and they naturally enforce multi-view coherency as inpainting is computed in a shared rectified texture space, allowing us to correctly respect perspective. We demonstrate instance-based object removal at the scale of a street in scenes composed of several hundreds of images. We next address outdoor relighting with a learning-based algorithm that efficiently allows the illumination in a scene to be changed, while removing and synthesizing cast shadows for any given sun position and accounting for global illumination. An approximate geometric proxy built using multi-view stereo is used to generate illumination and shadow related image buffers that guide a neural network. We train this network on a set of synthetic scenes allowing full supervision of the learning pipeline. Careful data augmentation allows our network to transfer to real scenes and provides state of the art relighting results. We also demonstrate the capacity of this network to be used to compose real scenes captured under different lighting conditions and orientation. We then present contributions to image-based rendering quality. We discuss how our carefully designed depth-map meshing and simplification algorithm improve rendering performance and quality of a new learning-based IBR method. Finally, we present a method that combines relighting, IBR, and material analysis. To enable relightable IBR with accurate glossy effects, we extract both material appearance variations and qualitative texture information from multi-view content in the form of several IBR heuristics. We further combine them with path-traced irradiance images that specify the input and target lighting. This combination allows a neural network to be trained to implicitly extract material properties and produce realistic-looking relit viewpoints. Separating diffuse and specular supervision is crucial in obtaining high-quality output
Nguyen, Tung Lam. "Contrôle et optimisation distribués basés sur l'agent dans les micro-réseaux avec implémentation Hardware-in-the-Loop". Thesis, Université Grenoble Alpes (ComUE), 2019. http://www.theses.fr/2019GREAT022/document.
Testo completoIn terms of the control hierarchy of microgrids, the coordination of local controllers is mandatory in the secondary and tertiary levels. Instead of using a central unit as conventional approaches, in this work, distributed schemes are considered. The distributed approaches have been taken attention widely recently due to the advantages of reliability, scalability, and security. The multi-agent system is an advanced technique having properties that make them suitable for acting as a basis for building modern distributed control systems. The thesis focuses on the design of agents aiming to distributed control and optimization algorithms in microgrids with realistic on-line deployment on a Hardware-in-the-loop platform. Based on the provided three-layer architecture of microgrids, a laboratory platform with Hardware-in-the-loop setup is constructed in the system level. This platform includes two parts: (1) a digital real-time simulator uses to simulate test case microgrids with local controllers in real-time; and (2) a cluster of hardware Raspberry PIs represents the multi-agent system operating in a sparse physical communication network. An agent is a Python-based program run on a single Raspberry PI owing abilities to transfer data with neighbors and computing algorithms to control the microgrid in a distributed manner.In the thesis, we apply the distributed algorithms for both secondary and tertiary control level. The distributed secondary controls in an islanded microgrid are presented in two approaches of finite-time consensus algorithm and average consensus algorithm with the improvements in performances. An extension of the platform with Power Hardware-in-the-Loop and IEC 61850-based communication is processed to make the deployment of agents closer to industrial applications. On the top control level, the agents execute the Alternating Direction Method of Multipliers to find out the optimal operation points of microgrid systems in both islanded and grid-connect state. The secondary and tertiary control objectives are achieved in a single framework which is rarely reported in other studies.Overall, the agent is explicitly investigated and deployed in the realistic conditions to facilitate applications of the distributed algorithms for the hierarchical control in microgrids. This research gives a further step making the distributed algorithms closer to onsite implementation
Ben, Dhaou Moaiz. "Optimisation du placement de tâches dans les systèmes distribués et de l'allocation de ressources pour les communications multipoints". Paris 11, 2003. http://www.theses.fr/2003PA112282.
Testo completoCombinatorial optimization problems consist in finding an optimal solution among a great number of feasible solutions. This kind of problems arise in many fields. In this thesis, we have studied optimization problems occuring in distributed systems and telecommunication networks. The problems we have considered are NP-complete, so there might not exist polynomial time algorithms to solve them optimally. We have, on the one hand, explored theoritical aspects of the problems: we studied some variants that can be solved either optimally or approximately with polynomial time algorithms. On the other hand, we looked at the practical side and proposed efficient heuristics. We first looked at a task assignment problem. For a given application, we seek to assign its tasks among different processors of a parallel architecture so as to minimize its overall completion time. We have proposed polynomial time approximation schemes to solve variants of the problem. Moreover, we have described many new heuristics to solve the general problem and we have provided methods to calculate lower bounds. The second problem we looked at arises in the field of telecommunications. It is a resources allocation problem in a multipoint communications setting. This problem, relatively recent, arised from the necessity of managing new applications for instance video-conference, group communication or network games. We have studied two problems. The first deals with network design (synthesis) under least cost and multipoint communications demand satisfaction. In the second, we seek to maximize the number of simultaneous multipoint communications in a network with fixed link capacities
Deswarte, Raphaël. "Régression linéaire et apprentissage : contributions aux méthodes de régularisation et d’agrégation". Thesis, Université Paris-Saclay (ComUE), 2018. http://www.theses.fr/2018SACLX047/document.
Testo completoThis thesis tackles the topic of linear regression, within several frameworks, mainly linked to statistical learning. The first and second chapters present the context, the results and the mathematical tools of the manuscript. In the third chapter, we provide a way of building an optimal regularization function, improving for instance, in a theoretical way, the LASSO estimator. The fourth chapter presents, in the field of online convex optimization, speed-ups for a recent and promising algorithm, MetaGrad, and shows how to transfer its guarantees from a so-called “online deterministic setting" to a “stochastic batch setting". In the fifth chapter, we introduce a new method to forecast successive intervals by aggregating predictors, without intermediate feedback nor stochastic modeling. The sixth chapter applies several aggregation methods to an oil production dataset, forecasting short-term precise values and long-term intervals
Baskiotis, Nicolas. "Evaluation d'algorithmes pour et par l'apprentissage". Paris 11, 2008. http://www.theses.fr/2008PA112275.
Testo completoOne of the main concerns involving machine learning applications is the a priori choice of an algorithm likely to yield the best classification performances. Our work is inspired by research in combinatorial optimisation on the phase transition problem. We suggest a dual approach to the standard view on this problem through metaleraning. Our goal is to build a competence map according to descriptors on the problem space which enables to identify the regimes where the system’s performance are steady. We assess this approach on C 4. 5. The second part of our work deals with machine learning problems in software testing. More precisely, we study a statistical structural method of software testing that uses the sampling of the so-called feasible paths in the control graph of the problems that is to be tested. In certain cases, the portion of feasible paths is very low, which hinders this method. Our goal is to design a learning and optimisation system that yields a random generator biaised towards the feasible paths and that warrantees a suitable sampling of the target concept. We designed the MLST system that implements active learning in a graph. We tested our work on both real and artificial problems and showed that our system achieves significantly improvement regarding the coverage of target concepts with respect to the initial data
Samir, Sara. "Approches coopératives pour certaines classes de problèmes d'optimisation non convexe : Algorithmes parallèles / distribués et applications". Electronic Thesis or Diss., Université de Lorraine, 2020. http://www.theses.fr/2020LORR0039.
Testo completoIn this thesis, we are interested in developing new cooperative approaches for solving some classes of nonconvex problems which play a very important role to model real-world problems. To design the schemes of our approaches, we combine several algorithms which we call the component (participant) algorithms. The combination is mainly based on DC (Difference of Convex Functions) and DCA (DC Algorithm) with metaheuristics. To develop our solution methods, we use the paradigm of parallel and distributed programming. Therefore, each process deals with an algorithm and communicates with the others by calling the functions of the MPI (Message Passing Interface) library which is a communication protocol in parallel and distributed programming. Besides the introduction and conclusion, this thesis is composed of four chapters. Chapter 1 concerns the theoretical and algorithmic tools serving as a methodological basis for the following chapters. Chapter 2 is about the mixed binary linear programs. To solve these problems, we propose a cooperative approach between DCA and VNS (Variable Neighborhood Search). Since the scheme is constituted by two algorithms, we use the point to point communication between the processes. As an application, we adapt our scheme to solve the capacitated facility location problem. Concerning chapter 3, we study the class of binary quadratic problems. Regarding the solution methods, we develop a cooperation between DCA-like which is a new version of DCA and two other metaheuristics: GA (Genetic Algorithm) and MBO (Migrating Birds Optimization). The exchange of information between the processes is expressed by using collective communication's function. More precisely, we call a function which allows broadcasting information of a process to all the others at the same time. This cooperative approach is adapted to the quadratic assignment problem. In chapter 4, we solve the MSSC (Minimum-Sum-of-Squares Clustering) using two cooperative approaches. The first combines DCA, VNS, and TS (Tabu Search). As for the second, it combines the MBO with the other three algorithms cited before. In these two approaches, we use a function of communication that allows a process to access the memories of the others and save the information there without blocking the work of the receiving processes
Taton, Christophe. "Vers l’auto-optimisation dans les systèmes autonomes". Grenoble INPG, 2008. http://www.theses.fr/2008INPG0183.
Testo completoThe increasing complexity of computer systems makes their administration even more tedious and error-prone. A general approach to this problem consists in building autonomic systems that are able to manage themselves and to handle changes of their state and their environment. While energy becomes even more scarce and expensive, the optimization of computer systems is an essential management field to improve their performance and to reduce their energetic footprint. As huge energy consumers, current computer systems are usually statically configured and behave badly in response to changes of their environment, and especially to changes of their workload. Self-optimization appears as a promising approach to these problems as it endows these systems with the ability to improve their own performance in an autonomous manner. This thesis focuses on algorithms and techniques to implement self-optimized autonomic systems. We specifically study self-optimization algorithms that rely on dynamic system provisioning in order to improve their performance and their resources’ efficiency. In the context of the Jade prototype of a component-based autonomic management platform, we propose best-effort algorithms that improve the performance of the managed systems through dynamic adaptations of the systems in response to gradual or sudden changes of their workload. We show the efficiency of these algorithms on Internet services and on messages services submitted to changing workloads. Finally, in order to guarantee optimal performance, we propose an optimization policy relying on the modelling of the managed system so as to generate optimal configurations. This policy is evaluated on a monitoring service for distributed systems. The implementation of autonomic management policies raised a number of challenges: the system is required to support dynamic adaptions, to observe itself and to take actions on itself. We address these needs with the Oz programming language and its distributed platform Mozart to implement the FructOz framework dedicated to the construction and handling of complex dynamic and distributed architecture-based systems, and the LactOz library specialized in the querying and browsing of dynamic architectures. Combining FructOz and LactOz, we show how to build complex dynamic systems involving distributed deployments as well as high levels of synchronizations and parameters
Arres, Billel. "Optimisation des performances dans les entrepôts distribués avec Mapreduce : traitement des problèmes de partionnement et de distribution des données". Thesis, Lyon, 2016. http://www.theses.fr/2016LYSE2012.
Testo completoIn this manuscript, we addressed the problems of data partitioning and distribution for large scale data warehouses distributed with MapReduce. First, we address the problem of data distribution. In this case, we propose a strategy to optimize data placement on distributed systems, based on the collocation principle. The objective is to optimize queries performances through the definition of an intentional data distribution schema of data to reduce the amount of data transferred between nodes during treatments, specifically during MapReduce’s shuffling phase. Secondly, we propose a new approach to improve data partitioning and placement in distributed file systems, especially Hadoop-based systems, which is the standard implementation of the MapReduce paradigm. The aim is to overcome the default data partitioning and placement policies which does not take any relational data characteristics into account. Our proposal proceeds according to two steps. Based on queries workload, it defines an efficient partitioning schema. After that, the system defines a data distribution schema that meets the best user’s needs, and this, by collocating data blocks on the same or closest nodes. The objective in this case is to optimize queries execution and parallel processing performances, by improving data access. Our third proposal addresses the problem of the workload dynamicity, since users analytical needs evolve through time. In this case, we propose the use of multi-agents systems (MAS) as an extension of our data partitioning and placement approach. Through autonomy and self-control that characterize MAS, we developed a platform that defines automatically new distribution schemas, as new queries appends to the system, and apply a data rebalancing according to this new schema. This allows offloading the system administrator of the burden of managing load balance, besides improving queries performances by adopting careful data partitioning and placement policies. Finally, to validate our contributions we conduct a set of experiments to evaluate our different approaches proposed in this manuscript. We study the impact of an intentional data partitioning and distribution on data warehouse loading phase, the execution of analytical queries, OLAP cubes construction, as well as load balancing. We also defined a cost model that allowed us to evaluate and validate the partitioning strategy proposed in this work
Moujahed, Sana. "Approche multi-agents réactive pour l'optimisation de systèmes spatialement distribués et dynamiques : application aux problèmes de positionnement". Besançon, 2007. http://www.theses.fr/2007BESA2019.
Testo completoThe work presented in this PhD thesis promotes the idea that the resolution of complex problems can be tackled thanks to a population of simple interacting agents. The objective of this thesis is to propose a self-organized approach to solve the single and multi-level facility location problem. This kind of problem requires locating facilies considering a certain demand, in order to optimize some performance criteria. The proposed model relies on a set-organizing simple agents situated in a common environment which interact and attempt to reach a global optimization goal. The agents have neither cognitive abilites nor a representation of the global system. The interactions between agents and their environment, which are based on the artificial potential fields approach, allow to locally optimize the agent's locations. In particular, the agents' behaviors are based on a combination of attractive and repulsive forces. The facility agents are attracted to the demand to satisfy their local objectives and repulsed by each other to ensure a consistent repartition in the environment. The optimization of the whole system is the outcome of a process of agents' self-organization. Our work has several concerns : agentifying the location problem, defining the solving process, and evaluating the approach relying on qualitative and quantitative criteria. We conduct empirical studies on various case studies. These, allow to handle several variants of the location problem, especially the multi-level problem, and to check the relevance of our approach
Girinon, Sylvain. "Étude de la stabilité et de la qualité des réseaux distribués de puissance". Thesis, Toulouse, INPT, 2010. http://www.theses.fr/2010INPT0010/document.
Testo completoThe emergence and the development of electrical systems during these last twenty years have led us to the elaboration of more and more complex architectures. They can be particularly found on embedded applications as well as in the heart of isolated distribution networks. The integration of several equipments with various natures raises the problem of stability. Thesis work presented here fits in with this context, leading to the implementation of stability and quality analysis methods, applied to distributed power networks. Studies led during this work are based on analytical expressions representing the continuous networks frequency behaviour. These models are then associated to the Routh-Hurwitz criterion in order to allow stability studies, according to their parameter values evolution. Analysis of results obtained on networks architectures using several equipments allows the refinement of our knowledge on these systems operation. Coupling phenomena, network layout according the loads number and power from a stability point of view, are particularly developed. Optimal sizing research for several undetermined elements, merging stability and quality criteria and carried out using optimization algorithms, is also presented. Finally, fundamental parts of this work which correspond to models building as well as stability studies are validated by an experimental approach
Montalbano, Pierre. "Contraintes linéaires et apprentissage sans conflit pour les modèles graphiques". Electronic Thesis or Diss., Toulouse 3, 2023. http://www.theses.fr/2023TOU30340.
Testo completoGraphical models define a family of formalisms and algorithms used in particular for logical and probabilistic reasoning, in fields as varied as image analysis or natural language processing. They are capable of being learned from data, giving probabilistic information that can then be combined with logical information. The goal of the thesis is to improve the efficiency of reasoning algorithms on these models crossing probabilities and logic by generalizing a fundamental mechanism of the most efficient purely logical reasoning tools (SAT solvers) to this hybrid case mixing probabilities and logic: conflict-based learning. The work is based on the concept of duality in linear programming and our learning mechanism is conflict-free, producing linear constraints efficiently solved using a knapsack formulation
Virouleau, Alain. "Apprentissage statistique pour la détection de données aberrantes et application en santé". Thesis, Institut polytechnique de Paris, 2020. http://www.theses.fr/2020IPPAX028.
Testo completoThe problems of outliers detection and robust regression in a high-dimensional setting are fundamental in statistics, and have numerous applications.Following a recent set of works providing methods for simultaneous robust regression and outliers detection,we consider in a first part a model of linear regression with individual intercepts, in a high-dimensional setting.We introduce a new procedure for simultaneous estimation of the linear regression coefficients and intercepts, using two dedicated sorted-l1 convex penalizations, also called SLOPE.We develop a complete theory for this problem: first, we provide sharp upper bounds on the statistical estimation error of both the vector of individual intercepts and regression coefficients.Second, we give an asymptotic control on the False Discovery Rate (FDR) and statistical power for support selection of the individual intercepts.Numerical illustrations, with a comparison to recent alternative approaches, are provided on both simulated and several real-world datasets.Our second part is motivated by a genetic problem. Among some particular DNA sequences called multi-satellites, which are indicators of the development or colorectal cancer tumors, we want to find the sequences that have a much higher (resp. much lower) rate of mutation than expected by biologist experts. This problem leads to a non-linear probabilistic model and thus goes beyond the scope of the first part. In this second part we thus consider some generalized linear models with individual intercepts added to the linear predictor, and explore the statistical properties of a new procedure for simultaneous estimation of the regression coefficients and intercepts, using again the sorted-l1 penalization. We focus in this part only on the low-dimensional case and are again interested in the performance of our procedure in terms of statistical estimation error and FDR
Sautet, Bernard. "Spécification et conception d'un outil de développement de systèmes temps-réel-strict distribués". Toulouse 3, 1996. http://www.theses.fr/1996TOU30100.
Testo completoBou, Nader Rami. "Stratégies tarifaires en assurance automobile : optimisation et expérimentation". Thesis, Université Paris-Saclay (ComUE), 2016. http://www.theses.fr/2016SACLN079.
Testo completoThe motor insurance sector currently confronts regulatory, financial, behavioral and technological challenges. Under these circumstances, insurers must uphold in improving their pricing strategies. Two topics related to pricing innovation are discussed in this thesis. We first take up the pricing strategy optimization for new businesses, as well as the renewals. Secondly, we highlight in the usage of experiments in leading us to a better understanding of insurance demand factors.On the first part of this thesis, we address pricing optimization at renewal, then illustrate how empirical demand models that rely on observable data could help the insurers to boost their profits and clients retention rate. We extend afterwards this framework by considering the impact of current pricing decisions on future cash-flows. Consequently, we introduce the Customer Value metric which allows insurers to reflect over the customers' behavior during their lifetime, when it comes to constructing their pricing strategy. The empirical illustrations of the first two chapters rely on natural data observed by the insurer.On the second part of this thesis, field and laboratory experiments will give us better comprehension of the motor insurance demand. Data from a field experiment refine the measure of clients' price elasticity. Offline assessment of several reinforcement learning algorithms shows how pricing experiments can achieve better performances compared with the myopic strategy which does not apply any kind of experiment. Laboratory experiments contribute to the understanding of demand models as well. In particular, we analyze the added value of risk aversion and risk perception in explaining the insurance choices. Furthermore, we examine the external validity of the experiment, i.e. the similarity between the behaviors of the customers in a lab environment versus their factual behaviors in the market.Aside from the duality between experiments and optimization, this thesis also illustrates the duality between private and public data, as well as the duality between empirical and theoretical insurance demand model
Tavakoli, Amirhossein. "Algorithmes hybrides d'optimisation combinatoire et d'apprentissage automatique pour l'efficacité énergétique des réseaux d'eau potable". Electronic Thesis or Diss., Université Côte d'Azur, 2023. http://www.theses.fr/2023COAZ4121.
Testo completoDrinking water distribution networks are energy-intensive systems, mainly due to pumping. However, they offer opportunities for load reduction and shifting, thanks to water towers and their storage capacity. Optimized operational management of pumping and storage in water networks, also known as “pump scheduling”, is therefore an advantageous lever for electricity networks, but it is also a complex mathematical optimization problem. The object of this thesis is the design of efficient resolution algorithms for a detailed, discrete and non-convex mathematical model. Unlike most of the literature on the subject, the emphasis is placed on the calculation of strictly feasible, possibly optimal, solutions of the mathematical model. Furthermore, the study strives to mitigate the algorithmic complexity of the problem due specifically to the coupling storage constraints, and presents different approaches to operate and exploit the temporal and spatial decomposition of the model.A first contribution thus concerns the design of preprocessing techniques for bound tightening and cut generation. Bounds and cuts are obtained from detailed partial (on a truncated time horizon or a part of the network) relaxations, and make it possible to reinforce a simpler (continuous and linear) general relaxation, basis of a global optimization algorithm. A second contribution concerns the development of an original local optimization algorithm, of the “Alternating Direction Method” type, which progressively refines a storage profile until reaching the associated valid pump schedule. Indeed, by fixing the coupling storage constraints at each iteration, the restricted non-convex discrete model can be solved exactly, by temporal and spatial decomposition. The algorithm thus recovers a feasible solution (a pumping plan) from a near-feasible near-optimal solution (a storage profile). If many heuristics from the literature can be invoked to generate this initial solution, we propose to obtain it by building a data model. The third contribution of the thesis thus concerns the development of a deep learning model, relying on the history of operations of a given network, and resulting in a data model complementary to the hybridized mathematical model. Scalability is an original feature of the approach, making it possible to learn a solution with a fine temporal discretization from a dataset for a coarse temporal discretization, thus remedying the difficulty of dataset generation. Finally, note that this hybrid combinatorial optimization and machine learning algorithm applies to other discrete optimal control problems of systems with storage. The empirical evaluation went through the generation of extensive training and experimentation datasets on networks from the literature and highlighted the performance of the exact and approximate algorithms
Canon, Louis-claude. "Outils et algorithmes pour gérer l'incertitude lors de l'ordonnancement d'application sur plateformes distribuées". Electronic Thesis or Diss., Nancy 1, 2010. http://www.theses.fr/2010NAN10097.
Testo completoThis thesis consists in revisiting traditional scheduling problematics in computational environments, and considering the adjunction of uncertainty in the models. We adopt here a wide definition of uncertainty that encompasses the intrinsic stochastic nature of some phenomena (e.g., processor failures that follow a Poissonian distribution) and the imperfection of model characteristics (e.g., inaccuracy of the costs in a model due to a bias in measurements). We also consider uncertainties that stem from indeterminations such as the user behaviors that are uncontrolled although being deterministic. Scheduling, in its general form, is the operation that assigns requests to resources in some specific way. In distributed environments, we are concerned by a workload (i.e., a set of tasks) that needs to be executed onto a computational platform (i.e., a set of processors). Therefore, our objective is to specify how tasks are mapped onto processors. Produced schedules can be evaluated through many different metrics (e.g., processing time of the workload, resource usage, etc) and finding an optimal schedule relatively to some metric constitutes a challenging issue. Probabilistic tools and multi-objectives optimization techniques are first proposed for tackling new metrics that arise from the uncertainty. In a second part, we study several uncertainty-related criteria such as the robustness (stability in presence of input variations) or the reliability (probability of success) of a schedule
Tremblet, David. "Apprentissage de contraintes pour améliorer la précision des modèles de planification et ordonnancement". Electronic Thesis or Diss., Ecole nationale supérieure Mines-Télécom Atlantique Bretagne Pays de la Loire, 2024. http://www.theses.fr/2024IMTA0417.
Testo completoManufacturing decisions often rely on mathematical models to suggest decisions to the managers in charge of production. For example, lot-sizing models are commonly used to plan factory production. The model calculates capacity usage for a plan with a rough approximation that does not account for all the complexities encountered on the shop floor. Although this approximation allows the model to be solved efficiently, the resulting decision usually leads to errors when the plan is executed on the shop floor. This thesis aims to use machine learning to improve the models traditionally used in operations research for manufacturing applications. The methodology aims to replace parts of the optimization models (constraints, objectives) with machine learning models (linear regression, neural networks, etc.) previously trained on available data. As a result, these tools can take advantage of the massive amount of data generated on the shop floor and external data sources to make better decisions. This approach is evaluated on a lot-sizing model where we learn capacity utilization constraints from the production schedule using machine learning models. The resulting model determines optimal production plans where production quantities remain feasible once sent to the shop floor. The resulting tool is also well adapted to today's production systems, which are increasingly reconfigurable and constantly evolving. The model can be retrained from shop floor data as changes occur on the shop floor, eliminating the need for an optimization expert to modify the optimization model each time the shop floor evolves
Dubois, Amaury. "Optimisation et apprentissage de modèles biologiques : application à lirrigation [sic l'irrigation] de pomme de terre". Thesis, Littoral, 2020. http://www.theses.fr/2020DUNK0560.
Testo completoThe subject of this PhD concerns one of the LISIC themes : modelling and simulation of complex systems, as well as optimization and automatic learning for agronomy. The objectives of the thesis are to answer the questions of irrigation management of the potato crop and the development of decision support tools for farmers. The choice of this crop is motivated by its important share in the Haut-de-France region. The manuscript is divided into 3 parts. The first part deals with continuous multimodal optimization in a black box context. This is followed by a presentation of a methodology for the automatic calibration of biological model parameters through reformulation into a black box multimodal optimization problem. The relevance of the use of inverse analysis as a methodology for automatic parameterisation of large models in then demonstrated. The second part presents 2 new algorithms, UCB Random with Decreasing Step-size and UCT Random with Decreasing Step-size. Thes algorithms are designed for continuous multimodal black-box optimization whose choice of the position of the initial local search is assisted by a reinforcement learning algorithms. The results show that these algorithms have better performance than (Quasi) Random with Decreasing Step-size algorithms. Finally, the last part focuses on machine learning principles and methods. A reformulation of the problem of predicting soil water content at one-week intervals into a supervised learning problem has enabled the development of a new decision support tool to respond to the problem of crop management
Rolet, Philippe. "Éléments pour l'Apprentissage et l'Optimisation de Fonctions Chères". Phd thesis, Université Paris Sud - Paris XI, 2010. http://tel.archives-ouvertes.fr/tel-00551865.
Testo completoNguyen, Manh Cuong. "La programmation DC et DCA pour certaines classes de problèmes en apprentissage et fouille de donées [i.e. données]". Electronic Thesis or Diss., Université de Lorraine, 2014. http://www.theses.fr/2014LORR0080.
Testo completoClassification (supervised, unsupervised and semi-supervised) is one of important research topics of data mining which has many applications in various fields. In this thesis, we focus on developing optimization approaches for solving some classes of optimization problems in data classification. Firstly, for unsupervised learning, we considered and developed the algorithms for two well-known problems: the modularity maximization for community detection in complex networks and the data visualization problem with Self-Organizing Maps. Secondly, for semi-supervised learning, we investigated the effective algorithms to solve the feature selection problem in semi-supervised Support Vector Machine. Finally, for supervised learning, we are interested in the feature selection problem in multi-class Support Vector Machine. All of these problems are large-scale non-convex optimization problems. Our methods are based on DC Programming and DCA which are well-known as powerful tools in optimization. The considered problems were reformulated as the DC programs and then the DCA was used to obtain the solution. Also, taking into account the structure of considered problems, we can provide appropriate DC decompositions and the relevant choice strategy of initial points for DCA in order to improve its efficiency. All these proposed algorithms have been tested on the real-world datasets including biology, social networks and computer security
Lesfari, Hicham. "Fondements réseaux et l'IA". Electronic Thesis or Diss., Université Côte d'Azur, 2022. http://www.theses.fr/2022COAZ4056.
Testo completoThe field of Artificial Intelligence (AI) has brought a broad impact on today's society, leading to a gripping interaction between several scientific disciplines. In this respect, there has been a strong twofold interest across the literature.On the one hand, a growing trend in telecommunication networks consists in revisiting classic optimization problems using machine learning techniques in order to exploit their potential benefits. We focus on some challenges brought by the detection of anomalies in networks, and the allocation of resources within software-defined networking (SDN) and network function virtualization (NFV).On the other hand, a substantial effort has been devoted towards the theoretical understanding of the collective behavior of networks. We focus on some challenges brought by the study of majority dynamics within multi-agent systems, and the compression of artificial neural networks with the aim at increasing their efficiency.In this study, we contextualize the above focal points in the framework of investigating some foundations of networks; viewed through the lens of telecommunications networks and neural networks. We first focus our attention on developing graph similarity measures for network anomaly detection. Next, we study deterministic and stochastic majority dynamics in multi-agent systems. Then, we discuss the random subset sum problem in the context of neural network compression. Finally, we walk through some other miscellaneous problems
Nguyen, Manh Cuong. "La programmation DC et DCA pour certaines classes de problèmes en apprentissage et fouille de donées [i.e. données]". Thesis, Université de Lorraine, 2014. http://www.theses.fr/2014LORR0080/document.
Testo completoClassification (supervised, unsupervised and semi-supervised) is one of important research topics of data mining which has many applications in various fields. In this thesis, we focus on developing optimization approaches for solving some classes of optimization problems in data classification. Firstly, for unsupervised learning, we considered and developed the algorithms for two well-known problems: the modularity maximization for community detection in complex networks and the data visualization problem with Self-Organizing Maps. Secondly, for semi-supervised learning, we investigated the effective algorithms to solve the feature selection problem in semi-supervised Support Vector Machine. Finally, for supervised learning, we are interested in the feature selection problem in multi-class Support Vector Machine. All of these problems are large-scale non-convex optimization problems. Our methods are based on DC Programming and DCA which are well-known as powerful tools in optimization. The considered problems were reformulated as the DC programs and then the DCA was used to obtain the solution. Also, taking into account the structure of considered problems, we can provide appropriate DC decompositions and the relevant choice strategy of initial points for DCA in order to improve its efficiency. All these proposed algorithms have been tested on the real-world datasets including biology, social networks and computer security
Bouzid, Salah Eddine. "Optimisation multicritères des performances de réseau d’objets communicants par méta-heuristiques hybrides et apprentissage par renforcement". Thesis, Le Mans, 2020. http://cyberdoc-int.univ-lemans.fr/Theses/2020/2020LEMA1026.pdf.
Testo completoThe deployment of Communicating Things Networks (CTNs), with continuously increasing densities, needs to be optimal in terms of quality of service, energy consumption and lifetime. Determining the optimal placement of the nodes of these networks, relative to the different quality criteria, is an NP-Hard problem. Faced to this NP-Hardness, especially for indoor environments, existing approaches focus on the optimization of one single objective while neglecting the other criteria, or adopt an expensive manual solution. Finding new approaches to solve this problem is required. Accordingly, in this thesis, we propose a new approach which automatically generates the deployment that guarantees optimality in terms of performance and robustness related to possible topological failures and instabilities. The proposed approach is based, on the first hand, on the modeling of the deployment problem as a multi-objective optimization problem under constraints, and its resolution using a hybrid algorithm combining genetic multi-objective optimization with weighted sum optimization and on the other hand, the integration of reinforcement learning to guarantee the optimization of energy consumption and the extending the network lifetime. To apply this approach, two tools are developed. A first called MOONGA (Multi-Objective Optimization of wireless Network approach based on Genetic Algorithm) which automatically generates the placement of nodes while optimizing the metrics that define the QoS of the CTN: connectivity, m-connectivity, coverage, k-coverage, coverage redundancy and cost. MOONGA tool considers constraints related to the architecture of the deployment space, the network topology, the specifies of the application and the preferences of the network designer. The second optimization tool is named R2LTO (Reinforcement Learning for Life-Time Optimization), which is a new routing protocol for CTNs, based on distributed reinforcement learning that allows to determine the optimal rooting path in order to guarantee energy-efficiency and to extend the network lifetime while maintaining the required QoS
Vo, Xuan Thanh. "Apprentissage avec la parcimonie et sur des données incertaines par la programmation DC et DCA". Electronic Thesis or Diss., Université de Lorraine, 2015. http://www.theses.fr/2015LORR0193.
Testo completoIn this thesis, we focus on developing optimization approaches for solving some classes of optimization problems in sparsity and robust optimization for data uncertainty. Our methods are based on DC (Difference of Convex functions) programming and DCA (DC Algorithms) which are well-known as powerful tools in optimization. This thesis is composed of two parts: the first part concerns with sparsity while the second part deals with uncertainty. In the first part, a unified DC approximation approach to optimization problem involving the zero-norm in objective is thoroughly studied on both theoretical and computational aspects. We consider a common DC approximation of zero-norm that includes all standard sparse inducing penalty functions, and develop general DCA schemes that cover all standard algorithms in the field. Next, the thesis turns to the nonnegative matrix factorization (NMF) problem. We investigate the structure of the considered problem and provide appropriate DCA based algorithms. To enhance the performance of NMF, the sparse NMF formulations are proposed. Continuing this topic, we study the dictionary learning problem where sparse representation plays a crucial role. In the second part, we exploit robust optimization technique to deal with data uncertainty for two important problems in machine learning: feature selection in linear Support Vector Machines and clustering. In this context, individual data point is uncertain but varies in a bounded uncertainty set. Different models (box/spherical/ellipsoidal) related to uncertain data are studied. DCA based algorithms are developed to solve the robust problems
Chen, Xiao. "Contrôle et optimisation de la perception humaine sur les vêtements virtuels par évaluation sensorielle et apprentissage de données expérimentales". Thesis, Lille 1, 2015. http://www.theses.fr/2015LIL10019/document.
Testo completoUnder the exacerbated worldwide competition, the mass customization or personalization of products is now becoming an important strategy for companies to enhance the perceived value of their products. However, the current online customization experiences are not fully satisfying for consumers because the choices are mostly limited to colors and motifs. The sensory fields of products, particularly the material’s appearance and hand as well as the garment fit are barely concerned.In my PhD research project, we have proposed a new collaborative design platform. It permits merchants, designers and consumers to have a new experience during the development of highly valued personalized garments without extra industrial costs. The construction of this platform consists of several parts. At first, we have selected, through a sensory experiment, an appropriate 3D garment CAD software in terms of rending quality. Then we have proposed an active leaning-based experimental design in order to find the most appropriate values of the fabric technical parameters permitting to minimize the overall perceptual difference between real and virtual fabrics in static and dynamic scenarios. Afterwards, we have quantitatively characterized the human perception on virtual garment by using a number of normalized sensory descriptors. These descriptors involve not only the appearance and the hand of the fabric but also the garment fit. The corresponding sensory data have been collected through two sensory experiments respectively. By learning from the experimental data, two models have been established. The first model permits to characterize the relationship between the appearance and hand perception of virtual fabrics and corresponding technical parameters that constitute the inputs of the 3D garment CAD software. The second model concerns the relationship between virtual garment fit perception and the pattern design parameters. These two models constitute the main components of the collaborative design platform. Using this platform, we have realized a number of garments meeting consumer’s personalized perceptual requirements
Nasri, Ridha. "Paramétrage Dynamique et Optimisation Automatique des Réseaux Mobiles 3G et 3G+". Phd thesis, Université Pierre et Marie Curie - Paris VI, 2009. http://tel.archives-ouvertes.fr/tel-00494190.
Testo completoVanet, Emmanuelle. "Distribution de l'intelligence et approche hétérarchique des marchés de l'énergie distribués dans les Smart Grids". Thesis, Université Grenoble Alpes (ComUE), 2016. http://www.theses.fr/2016GREAT112/document.
Testo completoIn close relationship with the European project DREAM, this doctoral thesis focus on operational evolutions in tomorrow’s distribution networks wich will integrate a larger amount of distributed renewable resources. A centralized control of all the entities (from controllable loads to embedded generators) is overall optimal but complex and not so reliable. This study addresses the feasibility of a distributed control, autonomous, self-learning and real time operation of local resources and network’s components. The main concern to explore will be the creation of ad-hoc federations of agents that will flexibly adjust their hierarchy to current needs. These collaborative structures will use different coordination strategies ranging from market-based transactions, to balancing optimization market (ancillary services) and to local control (frequency control and self-healing)
Leconte, Mathieu. "Équilibrage de charge et répartition de ressources dans les grands systèmes distribués". Phd thesis, Telecom ParisTech, 2013. http://tel.archives-ouvertes.fr/tel-00933645.
Testo completoPaquier, Williams. "Apprentissage ouvert de représentations et de fonctionalités en robotique : analyse, modèles et implémentation". Toulouse 3, 2004. http://www.theses.fr/2004TOU30233.
Testo completoAutonomous acquisition of representations and functionalities by a machine address several theoretical questions. Today’s autonomous robots are developed around a set of functionalities. Their representations of the world are deduced from the analysis and modeling of a given problem, and are initially given by the developers. This limits the learning capabilities of robots. In this thesis, we propose an approach and a system able to build open-ended representation and functionalities. This system learns through its experimentations of the environment and aims to augment a value function. Its objective consists in acting to reactivate the representations it has already learnt to connote positively. An analysis of the generalization capabilities to produce appropriate actions enable define a minimal set of properties needed by such a system. The open-ended representation system is composed of a network of homogeneous processing units and is based on position coding. The meaning of a processing unit depends on its position in the global network. This representation system presents similarities with the principle of numeration by position. A representation is given by a set of active units. This system is implemented in a suite of software called NeuSter, which is able to simulate million unit networks with billions of connections on heterogeneous clusters of POSIX machines. .
Ho, Vinh Thanh. "Techniques avancées d'apprentissage automatique basées sur la programmation DC et DCA". Thesis, Université de Lorraine, 2017. http://www.theses.fr/2017LORR0289/document.
Testo completoIn this dissertation, we develop some advanced machine learning techniques in the framework of online learning and reinforcement learning (RL). The backbones of our approaches are DC (Difference of Convex functions) programming and DCA (DC Algorithm), and their online version that are best known as powerful nonsmooth, nonconvex optimization tools. This dissertation is composed of two parts: the first part studies some online machine learning techniques and the second part concerns RL in both batch and online modes. The first part includes two chapters corresponding to online classification (Chapter 2) and prediction with expert advice (Chapter 3). These two chapters mention a unified DC approximation approach to different online learning algorithms where the observed objective functions are 0-1 loss functions. We thoroughly study how to develop efficient online DCA algorithms in terms of theoretical and computational aspects. The second part consists of four chapters (Chapters 4, 5, 6, 7). After a brief introduction of RL and its related works in Chapter 4, Chapter 5 aims to provide effective RL techniques in batch mode based on DC programming and DCA. In particular, we first consider four different DC optimization formulations for which corresponding attractive DCA-based algorithms are developed, then carefully address the key issues of DCA, and finally, show the computational efficiency of these algorithms through various experiments. Continuing this study, in Chapter 6 we develop DCA-based RL techniques in online mode and propose their alternating versions. As an application, we tackle the stochastic shortest path (SSP) problem in Chapter 7. Especially, a particular class of SSP problems can be reformulated in two directions as a cardinality minimization formulation and an RL formulation. Firstly, the cardinality formulation involves the zero-norm in objective and the binary variables. We propose a DCA-based algorithm by exploiting a DC approximation approach for the zero-norm and an exact penalty technique for the binary variables. Secondly, we make use of the aforementioned DCA-based batch RL algorithm. All proposed algorithms are tested on some artificial road networks
Vo, Xuan Thanh. "Apprentissage avec la parcimonie et sur des données incertaines par la programmation DC et DCA". Thesis, Université de Lorraine, 2015. http://www.theses.fr/2015LORR0193/document.
Testo completoIn this thesis, we focus on developing optimization approaches for solving some classes of optimization problems in sparsity and robust optimization for data uncertainty. Our methods are based on DC (Difference of Convex functions) programming and DCA (DC Algorithms) which are well-known as powerful tools in optimization. This thesis is composed of two parts: the first part concerns with sparsity while the second part deals with uncertainty. In the first part, a unified DC approximation approach to optimization problem involving the zero-norm in objective is thoroughly studied on both theoretical and computational aspects. We consider a common DC approximation of zero-norm that includes all standard sparse inducing penalty functions, and develop general DCA schemes that cover all standard algorithms in the field. Next, the thesis turns to the nonnegative matrix factorization (NMF) problem. We investigate the structure of the considered problem and provide appropriate DCA based algorithms. To enhance the performance of NMF, the sparse NMF formulations are proposed. Continuing this topic, we study the dictionary learning problem where sparse representation plays a crucial role. In the second part, we exploit robust optimization technique to deal with data uncertainty for two important problems in machine learning: feature selection in linear Support Vector Machines and clustering. In this context, individual data point is uncertain but varies in a bounded uncertainty set. Different models (box/spherical/ellipsoidal) related to uncertain data are studied. DCA based algorithms are developed to solve the robust problems
Rigollet, Philippe. "Inégalités d'oracle, agrégation et adaptation". Phd thesis, Université Pierre et Marie Curie - Paris VI, 2006. http://tel.archives-ouvertes.fr/tel-00115494.
Testo completoLes travaux faisant l'objet de cette thèse présentent différentes utilisations des inégalités d'oracle, d'abord dans un cadre général d'agrégation puis dans des modèles statistiques plus particuliers, comme l'estimation de densité et la classification. Les résultats obtenus sont une palette non exhaustive mais représentative de l'utilisation des inégalités d'oracle en statistique mathématique.