Дисертації з теми "Sous-espace de réseau de neurones"
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-31 дисертацій для дослідження на тему "Sous-espace de réseau de neurones".
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.
Gaya, Jean-Baptiste. "Subspaces of Policies for Deep Reinforcement Learning." Electronic Thesis or Diss., Sorbonne université, 2024. http://www.theses.fr/2024SORUS075.
This work explores "Subspaces of Policies for Deep Reinforcement Learning," introducing an innovative approach to address adaptability and generalization challenges in deep reinforcement learning (RL). Situated within the broader context of the AI revolution, this research emphasizes the shift toward scalable and generalizable models in RL, inspired by advancements in deep learning architectures and methodologies. It identifies the limitations of current RL applications, particularly in achieving generalization across varied tasks and domains, proposing a paradigm shift towards adaptive methods.The research initially tackles zero-shot generalization, assessing deep RL's maturity in generalizing across unseen tasks without additional training. Through investigations into morphological generalization and multi-objective reinforcement learning (MORL), critical limitations in current methods are identified, and novel approaches to improve generalization capabilities are introduced. Notably, work on weight averaging in MORL presents a straightforward method for optimizing multiple objectives, showing promise for future exploration.The core contribution lies in developing a "Subspace of Policies" framework. This novel approach advocates for maintaining a dynamic landscape of solutions within a smaller parametric space, taking profit of neural network weight averaging. Functional diversity is achieved with minimal computational overhead through weight interpolation between neural network parameters. This methodology is explored through various experiments and settings, including few-shot adaptation and continual reinforcement learning, demonstrating its efficacy and potential for scalability and adaptability in complex RL tasks.The conclusion reflects on the research journey, emphasizing the implications of the "Subspaces of Policies" framework for future AI research. Several future directions are outlined, including enhancing the scalability of subspace methods, exploring their potential in decentralized settings, and addressing challenges in efficiency and interpretability. This foundational contribution to the field of RL paves the way for innovative solutions to long-standing challenges in adaptability and generalization, marking a significant step forward in the development of autonomous agents capable of navigating a wide array of tasks seamlessly
Kirchhofer, Simon. "Conception d'une prothèse bio-inspirée commandée par réseaux de neurones exploitant les signaux électromyographiques." Thesis, Université Clermont Auvergne (2017-2020), 2020. http://www.theses.fr/2020CLFAC058.
Research on upper-body prosthetic device is commonly divided in two categories: The prosthesis mechatronic conception and the human-machine interface dedicated to the control. This PhD thesis aims to bring together these two fields of research. The first step deals with control signals. Thus, a database containing electromyographic sequences and vision based joint coordinate measurements was created. Then, an artificial neural network achieves the motion estimation from electromyographic sequences. Accordingly, an under-actuated bio-inspired hand architecture is proposed to copy an organic hand motion while ensuring a grasping force distribution. This innovative approach allows to optimize the synergies imitation and proposes a control more intuitive for active prosthesis users
Tzvetanov, Tzvetomir. "Etude psychophysique et modélisation des traitements de bas niveau sous-tendant la vision des contours des objets." Phd thesis, Université Louis Pasteur - Strasbourg I, 2003. http://tel.archives-ouvertes.fr/tel-00004179.
Ouss, Etienne. "Caractérisation des décharges partielles et identification des défauts dans les PSEM sous haute tension continue." Thesis, Lyon, 2018. http://www.theses.fr/2018LYSEC024.
The framework of this thesis is the monitoring of High-Voltage, Direct Current (HVDC) Gas-Insulated Substations (GIS). The availability of these equipment is crucial for electrical networks operators. That is why they need a preventive diagnosis tool. The solution must be able to detect and identify the insulation defects, so that an appropriate maintenance can be planned. The last 40 years have seen Partial Discharges (PD) measurement become a classic monitoring tool for AC GIS. Unfortunately, there is a lack of scientific information about PD in HVDC GIS, and the known defect identification techniques are very specific to the AC environment. New techniques are thus needed in DC.This thesis aimed to characterize partial discharges in DC gas-insulated substations, and to develop an automatic defect identification tool. The first step of this work was the development of a partial discharge measuring bench. The complete study has been performed in a GIS section, so that the results can be directly applied to industrial equipment. Two kinds of defect have been investigated: protrusions on the high-voltage conductor, and free metallic particles. The influence of parameters such as gas nature and pressure, voltage level and polarity has been evaluated. First, PD have been measured in conformity with the IEC 60270 standard, and the relevance of this method in a DC environment has been evaluated. Then, other measuring chains have been used to improve the characterization of partial discharges: a steady-state current measurement, a high-frequency current measurement, a light measurement and a measurement of Ultra-High Frequency (UHF) waves. Finally, a relevant signature for defect identification has been designed and extracted from DP recordings. A database has been constituted, and an automated recognition algorithm has been implemented.The results show that the conventional PD measurement technique is not fully adapted to partial discharges detection in DC, corona discharges being the most problematic situation. Nevertheless, this method has brought enough information to start the characterization of PD. The limitations of the conventional method have been explained thanks to the results of the other measurements. These other experimental results have led to an actual improvement of the characterization of protrusion and particle-generated partial discharges. An effective automated defect classification solution has been implemented. The signature is derived from the q(Δt) diagram that has been extracted from the data obtained with the partial discharge conventional measurement. The identification algorithm has a neural network structure
Dariouchy, Abdelilah. "Utilisation des réseaux de neurones artificiels en diffusion acoustique et en agriculture sous serres." Le Havre, 2008. http://www.theses.fr/2008LEHA0002.
This study is devoted to the models development able to predict the reduced cut-off frequencies and the forms functions for submerged tubes in water and to predict the acoustic spectrum retrodiffused by two welded plates on the one hand, and on the other hand, to predict the time series of the internal temperature and the internal moisture of the tomato greenhouse in a semi-arid area. To validate our results, the representation time-frequency of Wigner-Ville is used to compare the form function calculated by the traditional analytical method and that predicted by ANN. The control of the ANN models allows us now to consider other applications according to the requests
Demartines, Pierre. "Analyse de données par réseaux de neurones auto-organisés." Grenoble INPG, 1994. http://www.theses.fr/1994INPG0129.
Incerti, Sébastien. "Mesure de la fonction de structure polarisée g1n du neutron par l'expérience E154 au SLAC." Phd thesis, Université Blaise Pascal - Clermont-Ferrand II, 1998. http://tel.archives-ouvertes.fr/tel-00002876.
Jouini, Manel. "Reconstruction des images couleur de l'eau sous les nuages : recours à des méthodes neuronales." Paris 6, 2011. http://www.theses.fr/2011PA066508.
Giovannini, Francesco. "Modélisation mathématique pour l'étude des oscillations neuronales dans des réseaux de mémoire hippocampiques pendant l'éveil et sous anesthésie générale." Thesis, Université de Lorraine, 2017. http://www.theses.fr/2017LORR0182/document.
Memory is commonly defined as the ability to encode, store, and recall information we perceived. As we experience the world, we sense stimuli, we witness events, we ascertain facts, we study concepts, and we acquire skills. Although memory is an innate and familiar human behaviour, the interior workings of the brain which provide us with such faculties are far from being fully unravelled. Experimental studies have shown that during memory tasks, certain brain structures exhibit synchronous activity which is thought to be correlated with the short-term maintenance of salient stimuli. The objective of this thesis is to use biologically-inspired mathematical modelling and simulations of neural activity to shed some light on the mechanisms enabling the emergence of these memory-related synchronous oscillations. We focus in particular on hippocampal mnemonic activity during the awake state, and the amnesia and paradoxical memory consolidation occurring under general anaesthesia. We begin by introducing a detailed model of a type of persistent-firing pyramidal neuron commonly found in the CA3 and CA1 areas of the hippocampus. Stimulated with a brief transient current pulse, the neuron displays persistent activity maintained solely by cholinergic calcium-activated non-specific (CAN) receptors, and outlasting the stimulus for long delay periods (> 30s). Our model neuron and its parameters are derived from experimental in-vitro recordings of persistent firing hippocampal neurons carried out by our collaborators Beate Knauer and Motoharu Yoshida at the Ruhr University in Bochum, Germany. Subsequently, we turn our attention to the dynamics of a population of such interconnected pyramidal-CAN neurons. We hypothesise that networks of persistent firing neurons could provide the neural mechanism for the maintenance of memory-related hippocampal oscillations. The firing patterns elicited by this network are in accord with both experimental recordings and modelling studies. In addition, the network displays self-sustained oscillatory activity in the theta frequency. When connecting the pyramidal-CAN network to fast-spiking inhibitory interneurons, the dynamics of the model reveal that feedback inhibition improves the robustness of fast theta oscillations, by tightening the synchronisation of the pyramidal CAN neurons. We demonstrate that, in the model, the frequency and spectral power of the oscillations are modulated solely by the cholinergic mechanisms mediating the intrinsic persistent firing, allowing for a wide range of oscillation rates within the theta band. This is a biologically plausible mechanism for the maintenance of synchronous theta oscillations in the hippocampus which aims at extending the traditional models of septum-driven hippocampal rhythmic activity. In addition, we study the disruptive effects of general anaesthesia on hippocampal gamma-frequency oscillations. We present an in-depth study of the action of anaesthesia on neural oscillations by introducing a new computational model which takes into account the four main effects of the anaesthetic agent propofol GABAergic hippocampal interneurons. Our results indicate that propofol-mediated tonic inhibition contributes to enhancing network synchronisation in a network of hippocampal interneurons. This enhanced synchronisation could provide a possible mechanism supporting the occurrence of intraoperative awareness, explicit memory formation, and even paradoxical excitation under general anaesthesia, by facilitating the communication between brain structures which should supposedly be not allowed to do so when anaesthetised. In conclusion, the findings described within this thesis provide new insights into the mechanisms underlying mnemonic neural activity, both during wake and anaesthesia, opening compelling avenues for future work on clinical applications tackling neurodegenerative memory diseases, and anaesthesia monitoring
Giovannini, Francesco. "Modélisation mathématique pour l'étude des oscillations neuronales dans des réseaux de mémoire hippocampiques pendant l'éveil et sous anesthésie générale." Electronic Thesis or Diss., Université de Lorraine, 2017. http://www.theses.fr/2017LORR0182.
Memory is commonly defined as the ability to encode, store, and recall information we perceived. As we experience the world, we sense stimuli, we witness events, we ascertain facts, we study concepts, and we acquire skills. Although memory is an innate and familiar human behaviour, the interior workings of the brain which provide us with such faculties are far from being fully unravelled. Experimental studies have shown that during memory tasks, certain brain structures exhibit synchronous activity which is thought to be correlated with the short-term maintenance of salient stimuli. The objective of this thesis is to use biologically-inspired mathematical modelling and simulations of neural activity to shed some light on the mechanisms enabling the emergence of these memory-related synchronous oscillations. We focus in particular on hippocampal mnemonic activity during the awake state, and the amnesia and paradoxical memory consolidation occurring under general anaesthesia. We begin by introducing a detailed model of a type of persistent-firing pyramidal neuron commonly found in the CA3 and CA1 areas of the hippocampus. Stimulated with a brief transient current pulse, the neuron displays persistent activity maintained solely by cholinergic calcium-activated non-specific (CAN) receptors, and outlasting the stimulus for long delay periods (> 30s). Our model neuron and its parameters are derived from experimental in-vitro recordings of persistent firing hippocampal neurons carried out by our collaborators Beate Knauer and Motoharu Yoshida at the Ruhr University in Bochum, Germany. Subsequently, we turn our attention to the dynamics of a population of such interconnected pyramidal-CAN neurons. We hypothesise that networks of persistent firing neurons could provide the neural mechanism for the maintenance of memory-related hippocampal oscillations. The firing patterns elicited by this network are in accord with both experimental recordings and modelling studies. In addition, the network displays self-sustained oscillatory activity in the theta frequency. When connecting the pyramidal-CAN network to fast-spiking inhibitory interneurons, the dynamics of the model reveal that feedback inhibition improves the robustness of fast theta oscillations, by tightening the synchronisation of the pyramidal CAN neurons. We demonstrate that, in the model, the frequency and spectral power of the oscillations are modulated solely by the cholinergic mechanisms mediating the intrinsic persistent firing, allowing for a wide range of oscillation rates within the theta band. This is a biologically plausible mechanism for the maintenance of synchronous theta oscillations in the hippocampus which aims at extending the traditional models of septum-driven hippocampal rhythmic activity. In addition, we study the disruptive effects of general anaesthesia on hippocampal gamma-frequency oscillations. We present an in-depth study of the action of anaesthesia on neural oscillations by introducing a new computational model which takes into account the four main effects of the anaesthetic agent propofol GABAergic hippocampal interneurons. Our results indicate that propofol-mediated tonic inhibition contributes to enhancing network synchronisation in a network of hippocampal interneurons. This enhanced synchronisation could provide a possible mechanism supporting the occurrence of intraoperative awareness, explicit memory formation, and even paradoxical excitation under general anaesthesia, by facilitating the communication between brain structures which should supposedly be not allowed to do so when anaesthetised. In conclusion, the findings described within this thesis provide new insights into the mechanisms underlying mnemonic neural activity, both during wake and anaesthesia, opening compelling avenues for future work on clinical applications tackling neurodegenerative memory diseases, and anaesthesia monitoring
Baëtens, Tiffany. "Développement d'un microsystème étirable pour l'étude de l'activité électrique de cellules neuronales sous contrainte mécanique." Thesis, Lille 1, 2019. http://www.theses.fr/2019LIL1I100.
50 million people suffer from head trauma each year worldwide. The shock (external mechanical stress) acting on neural networks insided the brain can lead to the appearance of traumatic brain injuries (TBI). In the short term and long term, this process can lead to neuroinflammation and the manifestation of pathologies such as Parkinson’s disease or Alzheimer’s disease. This thesis aims to develop a stretchable microsystem for the electromechanical study of neural networks in vitro. Such a microsystem must integrate insulated electrodes which are mechanically robust on the stretchable support. The first part of this work consisted in studying the direct metallization on PDMS using physical masking. Next, a patterned photoresist (SU-8) thin film was then used between the PDMS and the metals. The stiff SU-8 shields the metallization from strain—which now occurs in the adjacent PDMS. In order to avoid cracks under mechanical stress, the architecture and orientation of the electrodes has been studied and validated by a physical model. The second part was the fabrication of a PDMS/SU-8/Cr-Au/Parylene stretchable microsystem by photolithography using a planar process developed during the thesis work. The microsystems is compatible with MultiChannel System© commercial device which allows the visualization and recording the nerve impulses of a mature neural network. A microfabrication process has been developed with 3 photolithography steps on a PDMS substrate. The functionality of the microsystem has been validated by the visualization of electrical responses of neural network at 12 days in vitro (DIV). In addition, this microsystem demonstrates a signal/noise ratio comparable to commercial MultiElectrode Array (MEA)
Le, Hai Son. "Continuous space models with neural networks in natural language processing." Phd thesis, Université Paris Sud - Paris XI, 2012. http://tel.archives-ouvertes.fr/tel-00776704.
Gallot, Guillaume. "Modélisation Dynamique et Commande d'un robot Anguille." Phd thesis, Ecole centrale de nantes - ECN, 2007. http://tel.archives-ouvertes.fr/tel-00306695.
A travers cet objectif, la thèse se porte dans un premier temps sur la modélisation dynamique du robot sous la forme d'un mécanisme hybride (structures de robots parallèles montées en série) permettant ainsi d'être le plus proche possible du prototype construit. Pour cela nous avons utilisé les algorithmes récursifs de Newton-Euler pour les modèles dynamiques inverse et direct en les généralisant au cas des robots à base mobile. Nous avons également proposé un modèle de contact fluide-structure pour simuler le comportement du robot dans l'eau. Pour tester ces algorithmes, nous avons simulé le comportement du robot lors de différents types de nage et en avons tiré des conclusions qui nous ont guidées dans la conception du prototype.
Dans un deuxième temps, à partir d'un générateur de mouvements à base de CPGs (ou réseau de neurones), nous avons étudié des lois de commande pour réaliser des simulations de nage en boucle fermée. Ainsi, nous avons abordé les problèmes de la nage vers des points cible et l'évitement d'obstacles pour la nage en milieu confiné.
Ammar, Karim. "Conception multi-physique et multi-objectif des cœurs de RNR-Na hétérogènes : développement d’une méthode d’optimisation sous incertitudes." Thesis, Paris 11, 2014. http://www.theses.fr/2014PA112390/document.
Since Phenix shutting down in 2010, CEA does not have Sodium Fast Reactor (SFR) in operating condition. According to global energetic challenge and fast reactor abilities, CEA launched a program of industrial demonstrator called ASTRID (Advanced Sodium Technological Reactor for Industrial Demonstration), a reactor with electric power capacity equal to 600MW. Objective of the prototype is, in first to be a response to environmental constraints, in second demonstrates the industrial viability of:• SFR reactor. The goal is to have a safety level at least equal to 3rd generation reactors. ASTRID design integrates Fukushima feedback;• Waste reprocessing (with minor actinide transmutation) and it linked industry.Installation safety is the priority. In all cases, no radionuclide should be released into environment. To achieve this objective, it is imperative to predict the impact of uncertainty sources on reactor behaviour. In this context, this thesis aims to develop new optimization methods for SFR cores. The goal is to improve the robustness and reliability of reactors in response to existing uncertainties. We will use ASTRID core as reference to estimate interest of new methods and tools developed.The impact of multi-Physics uncertainties in the calculation of the core performance and the use of optimization methods introduce new problems:• How to optimize “complex” cores (i.e. associated with design spaces of high dimensions with more than 20 variable parameters), taking into account the uncertainties?• What is uncertainties behaviour for optimization core compare to reference core?• Taking into account uncertainties, optimization core are they still competitive? Optimizations improvements are higher than uncertainty margins?The thesis helps to develop and implement methods necessary to take into account uncertainties in the new generation of simulation tools. Statistical methods to ensure consistency of complex multi-Physics simulation results are also detailed.By providing first images of innovative SFR core, this thesis presents methods and tools to reduce the uncertainties on some performance while optimizing them. These gains are achieved through the use of multi-Objective optimization algorithms. These methods provide all possible compromise between the different optimization criteria, such as the balance between economic performance and safety
Pouilly-Cathelain, Maxime. "Synthèse de correcteurs s’adaptant à des critères multiples de haut niveau par la commande prédictive et les réseaux de neurones." Electronic Thesis or Diss., université Paris-Saclay, 2020. http://www.theses.fr/2020UPASG019.
This PhD thesis deals with the control of nonlinear systems subject to nondifferentiable or nonconvex constraints. The objective is to design a control law considering any type of constraints that can be online evaluated.To achieve this goal, model predictive control has been used in addition to barrier functions included in the cost function. A gradient-free optimization algorithm has been used to solve this optimization problem. Besides, a cost function formulation has been proposed to ensure stability and robustness against disturbances for linear systems. The proof of stability is based on invariant sets and the Lyapunov theory.In the case of nonlinear systems, dynamic neural networks have been used as a predictor for model predictive control. Machine learning algorithms and the nonlinear observers required for the use of neural networks have been studied. Finally, our study has focused on improving neural network prediction in the presence of disturbances.The synthesis method presented in this work has been applied to obstacle avoidance by an autonomous vehicle
Compaore, Wendpuire Ousmane. "Aide à la décision pour le diagnostic des défauts pour une maintenance proactive dans un générateur photovoltaïque." Electronic Thesis or Diss., Normandie, 2023. http://www.theses.fr/2023NORMR095.
The loss of power of a photovoltaic generator (PVG) is undoubtedly due to the appearance of a certain number of anomalies linked to manufacturing, production or the environment and causing failures in its proper functioning. From a realistic model, quite close to real operation and able to take into account the avalanche effect of a PN junction transmitted to the entire PVG, we have sufficiently shown the loss of performance of a PV generator and the need to have a diagnostic method for maintenance assistance in order not to suffer the effects of faults.Two diagnostic methods were applied to this PVG, one relating to the detection and localization of sensor faults, and the other to the detection and localization of system faults. The particular choice of these two diagnostic techniques, which do not target the same types of faults, lies in the complex nature of the model of the industrial process subjected to study. The performances obtained with the analytical redundancy relations (ARR), method based on the principle of parity space applied to the maximum operating point are very relevant. Using the artificial intelligence (AI), method based on the principle of artificial neural networks (ANN), we experimented with two classification methods for the detection and diagnosis of system faults. If detectability is proven with our different configurations without the possibility of locating the origin and the cause in the first part of the classification, we arrive thanks to a bundle of clues to locate the origin or the cause thanks to the classification for the diagnostic.The production of two real-time acquisition prototypes is based on the principle of the Industrial Internet of Things (IIoT). The first only allows the acquisition and saving of data on an SD card. The second, and more advanced prototype, allows real-time transmission via WiFi to a web server and aims to create a real-time monitoring platform in the long term. Both prototypes produce data that is used to power both diagnostic methods. The results obtained with real data are compatible with those obtained in the simulation phase. The conclusions of this diagnosis will enable greater efficiency in proactive maintenance operations
Guerra, Jonathan. "Optimisation multi-objectif sous incertitudes de phénomènes de thermique transitoire." Thesis, Toulouse, ISAE, 2016. http://www.theses.fr/2016ESAE0024/document.
This work aims at solving multi-objective optimization problems in the presence of uncertainties and costly numerical simulations. A validation is carried out on a transient thermal test case. First of all, we develop a multi-objective optimization algorithm based on kriging and requiring few calls to the objective functions. This approach is adapted to the distribution of the computations and favors the restitution of a regular approximation of the complete Pareto front. The optimization problem under uncertainties is then studied by considering the worst-case and probabilistic robustness measures. The superquantile integrates every event on which the output value is between the quantile and the worst case. However, it requires an important number of calls to the uncertain objective function to be accurately evaluated. Few methods give the possibility to approach the superquantile of the output distribution of costly functions. To this end, we have developed an estimator based on importance sampling and kriging. It enables to approach superquantiles with little error and using a limited number of samples. Moreover, the setting up of a coupling with the multi-objective algorithm allows to reuse some of those evaluations. In the last part, we build spatio-temporal surrogate models capable of predicting non-linear, dynamic and long-term in time phenomena by using few learning trajectories. The construction is based on recurrent neural networks and a construction facilitating the learning is proposed
Joucla, Sébastien. "Etude expérimentale et modélisation de la stimulation électrique extracellulaire des réseaux de neurones avec des matrices de microélectrodes (MEA) : analyse des mécanismes sous-jacents et amélioration de la focalité spatiale des stimulations." Bordeaux 1, 2007. http://www.theses.fr/2007BOR13523.
Extracellular electrical stimulation of the central nervous system using microelectrode arrays (MEAs) is currently a challenging stake in neuroscience, in the fields of both fundamental and clinical research. This thesis aims at understanding the mechanisms underlying the extracellular stimulation of neurons. An experimental study shows that monopolar stimulations are not focal. Through a computational study, a finite element model was developed and used for realistic computation of the extracellular potential created by a stimulation. Based on this model, a new electrode configuration is proposed to achieve focal spatial stimulations (patent pending). Finally, a model made of a compartmentalized neuron placed in an extracellular potential field was built and used to explain several effects of extracellular stimulation on the neural response. This work allows a better control of extracellular stimulation for their use to study activity-dependent phenomena underlying neural network plasticity and for the development of efficient neural prostheses
Brozzoli, Claudio. "Peripersonal space : a multisensory interface for body-objects interactions." Phd thesis, Université Claude Bernard - Lyon I, 2009. http://tel.archives-ouvertes.fr/tel-00675247.
Boukhtache, Seyfeddine. "Système de traitement d’images temps réel dédié à la mesure de champs denses de déplacements et de déformations." Thesis, Université Clermont Auvergne (2017-2020), 2020. http://www.theses.fr/2020CLFAC054.
This PhD thesis has been carried out in a multidisciplinary context. It deals with the challenge of real-time and metrological performance in digital image processing. This is particularly interesting in photomechanics. This is a recent field of activity, which consists in developing and using systems for measuring whole fields of small displacements and small deformations of solids subjected to thermomechanical loading. The technique targeted in this PhD thesis is Digital Images Correlation (DIC), which is the most popular measuring technique in this community. However, it has some limitations, the main one being the computing resources and the metrological performance, which should be improved to reach that of classic pointwise measuring sensors such as strain gauges.In order to address this challenge, this work relies on two main studies. The first one consists in optimizing the interpolation process because this is the most expensive treatment in DIC. Acceleration is proposed by using a parallel hardware implementation on FPGA, and by taking into consideration the consumption of hardware resources as well as accuracy. The main conclusion of this study is that a single FPGA (current technology) is not sufficient to implement the entire DIC algorithm. Thus, a second study has been proposed. It is based on the use of convolutional neural networks (CNNs) in an attempt to achieve both better metrological performance than CIN and real-time processing. This second study shows the relevance of using CNNs for measuring displacement and deformation fields. It opens new perspectives in terms of metrological performance and speed of full-field measuring systems
Burnod, Yves. "Modèle de cortex cérébral et implémentation sur un réseau de processeurs parallèles." Angers, 1988. http://www.theses.fr/1988ANGE0005.
Vollant, Antoine. "Evaluation et développement de modèles sous-maille pour la simulation des grandes échelles du mélange turbulent basés sur l'estimation optimale et l'apprentissage supervisé." Thesis, Université Grenoble Alpes (ComUE), 2015. http://www.theses.fr/2015GREAI118/document.
This work develops subgrid model techniques and proposes methods of diagnosis for Large Eddy Simulation (LES) of turbulent mixing.Several models from these strategies are thus presented to illustrate these methods.The principle of LES is to solve the largest scales of the turbulent flow responsible for major transfers and to model the action of small scales of flowon the resolved scales. Formally, this operation leads to filter equations describing turbulent mixing. Subgrid terms then appear and must bemodeled to close the equations. In this work, we rely on the classification of subgrid models into two categories. "Functional" models whichreproduces the energy transfers between the resolved scales and modeled scales and "Structural" models that seek to reproduce the exact subgrid termitself. The first major challenge is to evaluate the performance of subgrid models taking into account their functional behavior (ability to reproduce theenergy transfers) and structural behaviour (ability to reproduce the term subgrid exactly). Diagnostics of subgrid models have been enabled with theuse of the optimal estimator theory which allows the potential of structural improvement of the model to be evaluated.These methods were initially involved for the development of a first family of models called algebraic subgrid $DRGM$ for "Dynamic Regularized GradientModel". This family of models is based on the structural diagnostic of terms given by the regularization of the gradient model family.According to the tests performed, this new structural model's family has better functional and structural performance than original model's family of thegradient. The improved functional performance is due to the vanishing of inverse energy transfer (backscatter) observed in models of thegradient family. This allows the removal of the unstable behavior typically observed for this family of models.In this work, we then propose the use of the optimal estimator directly as a subgrid scale model. Since the optimal estimator provides the modelwith the best structural performance for a given set of variables, we looked for the set of variables which optimize that performance. Since this set of variablesis large, we use surrogate functions of artificial neural networks type to estimate the optimal estimator. This leads to the "Artificial Neural Network Model"(ANNM). These alternative functions are built from databases in order to emulate the exact terms needed to determine the optimal estimator. The tests of this modelshow that he it has very good performance for simulation configurations not very far from its database used for learning, so these findings may fail thetest of universality.To overcome this difficulty, we propose a hybrid method using an algebraic model and a surrogate model based on artificial neural networks. Thebasis of this new model family $ACM$ for "Adaptive Coefficient Model" is based on vector and tensor decomposition of the exact subgrid terms. Thesedecompositions require the calculation of dynamic coefficients which are modeled by artificial neural networks. These networks have a learning method designedto directlyoptimize the structural and functional performances of $ACM$. These hybrids models combine the universality of algebraic model with high performance butvery specialized performance of surrogate models. The result give models which are more universal than ANNM
Pham, Hoang Anh. "Coordination de systèmes sous-marins autonomes basée sur une méthodologie intégrée dans un environnement Open-source." Electronic Thesis or Diss., Toulon, 2021. http://www.theses.fr/2021TOUL0020.
This thesis studies the coordination of autonomous underwater robots in the context of coastal seabed exploration or facility inspections. Investigating an integrated methodology, we have created a framework to design and simulate low-cost underwater robot controls with different model assumptions of increasing complexity (linear, non-linear, and finally non-linear with uncertainties). By using this framework, we have studied algorithms to solve the problem of formation control, collision avoidance between robots and obstacle avoidance of a group of underwater robots. More precisely, we first consider underwater robot models as linear systems of simple integrator type, from which we can build a formation controller using consensus and avoidance algorithms. We then extend these algorithms for the nonlinear dynamic model of a Bluerov robot in an iterative design process. Then we have integrated a Radial Basis Function neural network, already proven in convergence and stability, with the algebraic controller to estimate and compensate for uncertainties in the robot model. Finally, we have presented simulation results and real basin tests to validate the proposed concepts. This work also aims to convert a remotely operated ROV into an autonomous ROV-AUV hybrid
Blagouchine, Iaroslav. "Modélisation et analyse de la parole : Contrôle d’un robot parlant via un modèle interne optimal basé sur les réseaux de neurones artificiels. Outils statistiques en analyse de la parole." Thesis, Aix-Marseille 2, 2010. http://www.theses.fr/2010AIX26666.
This Ph.D. dissertation deals with speech modeling and processing, which both share the speech quality aspect. An optimum internal model with constraints is proposed and discussed for the control of a biomechanical speech robot based on the equilibrium point hypothesis (EPH, lambda-model). It is supposed that the robot internal space is composed of the motor commands lambda of the equilibrium point hypothesis. The main idea of the work is that the robot movements, and in particular the robot speech production, are carried out in such a way that, the length of the path, traveled in the internal space, is minimized under acoustical and mechanical constraints. Mathematical aspect of the problem leads to one of the problems of variational calculus, the so-called geodesic problem, whose exact analytical solution is quite complicated. By using some empirical findings, an approximate solution for the proposed optimum internal model is then developed and implemented. It gives interesting and challenging results, and shows that the proposed internal model is quite realistic; namely, some similarities are found between the robot speech and the real one. Next, by aiming to analyze speech signals, several methods of statistical speech signal processing are developed. They are based on higher-order statistics (namely, on normalized central moments and on the fourth-order cumulant), as well as on the discrete normalized entropy. In this framework, we also designed an unbiased and efficient estimator of the fourth-order cumulant in both batch and adaptive versions
Quininao, Cristobal. "Mathematical modeling in neuroscience : collective behavior of neuronal networks & the role of local homeoproteins diffusion in morphogenesis." Thesis, Paris 6, 2015. http://www.theses.fr/2015PA066152/document.
This work is devoted to the study of mathematical questions arising from the modeling of biological systems combining analytic and probabilistic tools. In the first part, we are interested in the derivation of the mean-field equations related to some neuronal networks, and in the study of the convergence to the equilibria of the solutions to the limit equations. In Chapter 2, we use the coupling method to prove the chaos propagation for a neuronal network with delays and random architecture. In Chapter 3, we consider a kinetic FitzHugh-Nagumo equation. We analyze the existence of solutions and prove the nonlinear exponential convergence in the weak connectivity regime. In the second part, we study the role of homeoproteins (HPs) on the robustness of boundaries of functional areas. In Chapter 4, we propose a general model for neuronal development. We prove that in the absence of diffusion, the HPs are expressed on irregular areas. But in presence of diffusion, even arbitrarily small, well defined boundaries emerge. In Chapter 5, we consider the general model in the one dimensional case and prove the existence of monotonic stationary solutions defining a unique intersection point for any arbitrarily small diffusion coefficient. Finally, in the third part, we study a subcritical Keller-Segel equation. We show the chaos propagation without any restriction on the force kernel. Eventually, we demonstrate that the propagation of chaos holds in the entropic sense
Swaileh, Wassim. "Des modèles de langage pour la reconnaissance de l'écriture manuscrite." Thesis, Normandie, 2017. http://www.theses.fr/2017NORMR024/document.
This thesis is about the design of a complete processing chain dedicated to unconstrained handwriting recognition. Three main difficulties are adressed: pre-processing, optical modeling and language modeling. The pre-processing stage is related to extracting properly the text lines to be recognized from the document image. An iterative text line segmentation method using oriented steerable filters was developed for this purpose. The difficulty in the optical modeling stage lies in style diversity of the handwriting scripts. Statistical optical models are traditionally used to tackle this problem such as Hidden Markov models (HMM-GMM) and more recently recurrent neural networks (BLSTM-CTC). Using BLSTM we achieve state of the art performance on the RIMES (for French) and IAM (for English) datasets. The language modeling stage implies the integration of a lexicon and a statistical language model to the recognition processing chain in order to constrain the recognition hypotheses to the most probable sequence of words (sentence) from the language point of view. The difficulty at this stage is related to the finding the optimal vocabulary with minimum Out-Of-Vocabulary words rate (OOV). Enhanced language modeling approaches has been introduced by using sub-lexical units made of syllables or multigrams. The sub-lexical units cover an important portion of the OOV words. Then the language coverage depends on the domain of the language model training corpus, thus the need to train the language model with in domain data. The recognition system performance with the sub-lexical units outperformes the traditional recognition systems that use words or characters language models, in case of high OOV rates. Otherwise equivalent performances are obtained with a compact sub-lexical language model. Thanks to the compact lexicon size of the sub-lexical units, a unified multilingual recognition system has been designed. The unified system performance have been evaluated on the RIMES and IAM datasets. The unified multilingual system shows enhanced recognition performance over the specialized systems, especially when a unified optical model is used
Lamouret, Marie. "Traitement automatisés des données acoustiques issues de sondeurs multifaisceaux pour la cartographie des fonds marins." Electronic Thesis or Diss., Toulon, 2022. http://www.theses.fr/2022TOUL0002.
Among underwater acoustic technologies, multibeam echo sounder (MBES) is one of the most advanced tool to study and map the underwater floors and the above water column. Its deployment on-site requires expertise so as the whole data processing to map the information. These processing are very time-consuming due to the massive quantity of recorded data and thus needs to be automatised to shorten and alleviate the hydrographer's task. This PhD research works focus on the automatisation of the current activities in Seaviews society.After some reminders on the underwater acoustic sciences, the MBES operating is described as well the produced data that will be manipulated throughout the developments. This document presents two thematics˸ bathymetric (depths) and marine habitats mapping. The developments are integrated into the Seaviews' software in the aim to be used by all the employees.About seafloor depths mapping, the bathymetric sounding has to be sorted to avoid that the outlier errors distort the results. Sorting the uncountable measures is cumbersome but necessary, although the hydrographers are today happily computed-assisted. We propose a fast statistical method to exclude the outliers while mapping the information. This leads to wonder if the water column imagery would be workable to deduce the bathymetry without failure. We will test this hypothesis with some technics of deep learning, especially with convolutional neural networks.The marine habitats mapping is a seabed nature classification according to the local life. Seaviews has worked on a way to prepare MBES data and habitats analysis. Concerning the method of classification itself, we move towards machine learning technics. Several methods are implemented and assessed, and then an area is chosen to evaluate and compare the results
Fitoussi, Aurélie. "Marqueurs comportementaux et corrélats neurobiologiques de la prise de décision adaptée et inadaptée chez le rat." Thesis, Bordeaux 2, 2011. http://www.theses.fr/2011BOR21871.
Decision-making is profoundly impaired in several psychiatric disorders such as addiction, but also in some healthy individuals for whom immediate gratifications prevail over long term gain. To better elucidate the neuropsychological and neurobiological bases of good and poor decision making in normal and pathological conditions, healthy poor decision-makers represent a promising model. Recently, a Rat Gambling Task, aimed at measuring decision-making like in the Iowa gambling Task in humans has been validated. It allows the identification, among a normal population of rats, of majority of good decision-makers, and a minority of poor decision-makers that prefer immediate larger reward despite suffering large loses. We demonstrated that all poor decision makers are unflexible and less efficient in goal-directed behavior. They also have a higher motivation for reward that depends on a complex cost/benefice balance, related to the effort to make, to food palatability, but not to the perception of the pleasant feeling or to metabolic needs. Moreover, we demonstrated the absence of relationship between decision making performance and working memory. At the neurobiological level, we demonstrated 1) that efficiency in goal-directed behavior depends on balance of activity between PL and SDM and 2) that decision making depends on specific brain regions, with a level of activity related to the performance, as well as the time course to make choices. Higher OFC and Nacc shell activities are systematically associated with good decision making, whereas the recruitment of PL/SDM is modulated according to the time course to make good choices. CgA, IL and the amygdala would be disengaged when choices are established. Poor decision makers display a prefrontal hypoactivity associated with a persistent involvement of the amygdala, suggesting an alteration in the prefrontal cognitive control, combined with deficits in reward-based associations, leading to an impaired acquisition and/or re-updating of the incentive value of the options. Moreover, we demonstrated that inter-individual differences in the RGT are associated with distinct DA- and 5HT basal functions. Poor decision makers notably displayed (1) high DA- and 5HT-ergic metabolisms in IL, supporting their motor impulsivity and/or lower efficiency in goal-directed behavior and (2) a higher DA-ergic metabolism in the Nacc core, and lower 5HT-ergic in BLA, that could be related to their higher motivation, and the quality of reward-based associations. These data support the relationship between genetic polymorphisms inducing distinct basal monoaminergic functioning, and poor decision making as well as psychiatric disorders. All these cognitive/behavioural and neurobiological characteristics that make a consistent framework could be an endophenotype of mental disorders. Further experiments should examine the direct relationship between poor decision making and psychiatric disorders, such as addiction, and the genetic background related to this specific profile
Zaidi, Donia. "Étude des mécanismes pathogéniques dépendants des microtubules dans les progéniteurs neuronaux conduisant aux malformations corticales." Electronic Thesis or Diss., Sorbonne université, 2023. https://accesdistant.sorbonne-universite.fr/login?url=https://theses-intra.sorbonne-universite.fr/2023SORUS159.pdf.
In mammals, cortical development is a finely regulated process that leads to the formation of a functional cortex. Apical radial glial cells (RG) are key progenitor cells du ring cortical development, capable of self-renewal or neuronal generation, with a soma restricted to the ventricular zone (VZ) in rodents. Their nucleus migrates according to the phases of the cell cycle by a process called interkinetic nuclear migration (INM). RG have a bipolar shape, with a long basal process supporting neuronal migration and a short apical process facing the ventricle where a primary cilium (PC), anchored to a modified centrosome (‘basal body’), emerges and detects molecules present in the embryonic cerebrospinal fluid. Genetic mutations can alter the function of RG, affecting cortical development and leading to cortical malformations. These malformations are associated in patients with epilepsy, intellectual disabilities and also neuropsychiatric disorders. It is therefore important to determine how the molecular and cellular processes involving RG can be disrupted by genetic mutations. Thus, my thesis work focused on the study of mutations affecting two different genes in the context of two rare cortical malformations. First, the gene encoding for the motor protein dynein heavy chain (DYNC1H1) was found mutated in patients with a complex cortical malformation associated with microcephaly (small brain) and dysgyria (gyri defects). We generated a Knock-In (KI) mouse model for this gene, reproducing a missense mutation found in a patient. During my thesis, I studied RG at mid-corticogenesis of this KI model and, by comparing it with a mouse model mutant for the same gene but leading to peripheral neuropathies, we showed RG alterations specific to the KI model. We found abnormalities in INM, cell cycle and neuronal migration. Also, defects of key organelles, such as mitochondria and Golgi apparatus were identified in progenitors and are specific in the cortical malformation KI model. Secondly, subcortical heterotopia (SH) is a cortical malformation characterized by the abnormal presence of neurons in the white matter. Mutations in the gene coding for EML1 (Echinoderm microtubule associated protein like 1) were identified in certain SH patients. When Eml1 is mutated in mice, numerous RG are found in basal positions of the cortical wall outside the VZ, suggesting that they detach apically. Within the apical process, abnormal PC formation and basal bodies were described. By studying a new mutant mouse model where Eml1 is inactivated, my work focused on subcellular and cellular alterations of RG to understand the pathogenic mechanisms leading to their detachment and thus to SH formation. In interphase RG, focusing on mechanisms upstream of PC formation, I analyzed centrosomes and determined that their structure is affected in patient and mouse mutant cells, and these defects are rescued by stabilizing microtubules. Recruitment of key centrosomal proteins is altered early in development, and the centrosomal protein Cep170 was found to be a specific interacting partner of EML1, this interaction being lost when EML1 carries a patient mutation. Because centrosomes and cilia are intimately linked to the cell cycle, I proceeded to analyze the RG cell cycle and identified alterations in cell cycle kinetics during early and mid-development. Single-cell RNA sequencing at two key developmental stages identified deregulations in cell cycle gene expression. Abnormal RG detachment appears greater in early compared to mid-development, suggesting that centrosomal and cell cycle alterations at this stage may be upstream of abnormal RG detachment. My thesis work thus brings new elements essential to the understanding of the altered mechanisms in neural progenitors related to rare cortical malformations
Saeed, Kashif. "CONTRIBUTION A LA SURVEILLANCE DE L'INTEGRITE DES STRUCTURES." Phd thesis, 2010. http://pastel.archives-ouvertes.fr/pastel-00503109.
Simard, Michel. "Mémoires de traduction sous-phrastiques." Thèse, 2003. http://hdl.handle.net/1866/14509.